We have studied the effect of common mutations (677C → T and 1298A → C) of the methylenetetrahydrofolate reductase (MTHFR) gene in sixty-six healthy French subjects, aged 27–47 years. Serum folate, vitamin B12, and plasma total homocysteine were measured as well as the specific activity of MTHFR in lymphocytes. The frequency of subjects homozygous for the 677TT genotype was 18 %, and that of those homozygous for the 1298CC genotype was 12·5 %. The frequency of individuals heterozygous for both mutations was 23·5 %. The 1298A → C mutation was associated with decreased MTHFR specific activity in subjects with both 677CC and 677CT genotypes. This activity was 60 % for the 677CC/1298AC genotype and 52 % for the 677CC/1298CC genotype when compared with the MTHFR specific activity of the 677CC/1298AA genotype. Heterozygotes for both mutations (677CT/1298AC genotype) had 36 % of the reference specific activity. Although homocysteine levels in 677TT and 1298CC genotype subjects were higher than for other genotypes, no significant differences were observed among different genotypes. This may be due to high serum folate level in our samples, and suggests that folate therapy may be useful to prevent hyperhomocysteinaemia in homozygous mutant subjects.

The objectives of the present study were to determine whether differences in usual physical activity affect BMR, sleeping energy expenditure (EE), and EE during seated activities between athletic and non-athletic adolescents, and to establish individual relationships between heart rate and EE. Adolescents (n 49, four groups of eleven to fifteen boys or girls aged 16–19 years) participated in the study. Body composition was measured by the skinfold-thickness method and maximum O2 consumption (VO2max) by a direct method (respiratory gas exchange) on a cycloergometer. The subjects each spent 36 h in one of two large whole-body calorimeters. They followed a standardized activity programme including two periods of exercise simulating their mean weekly physical activities. Fat-free mass (FFM), VO2max, daily EE and EE during sleep and seated activities were significantly higher in athletic than in non-athletic subjects of both sexes. VO2max, daily EE and EE during exercise adjusted for FFM were higher in athletic than in non-athletic adolescents (P < 0·001), whereas sleeping EE, BMR and EE during seated activities and adjusted for FFM were not significantly different between athletic and non-athletic adolescents. However, sex differences in EE remained significant. Thus, differences in EE between athletic and non-athletic adolescents resulted mainly from differences in FFM and physical exercise. Usual activity did not significantly affect energy utilization of substrates. Finally, individual relationships were computed between heart rate and EE with activity programmes simulating the usual activities of athletic and non-athletic adolescents with the goal of predicting EE of the same subjects in free-living conditions.

The main Fe storage organ in the body is the liver. In patients with chronic liver disease, secondary Fe overload is common. Phlebotomy, often used in the West to reduce Fe overload to improve the efficacy of interferon therapy, is not socially acceptable in India. We assessed the efficacy of a low-Fe diet in reducing serum Fe levels. Nineteen patients with hepatitis B- and C-related chronic liver disease, ten with normal (< 25 μmol/l) baseline serum Fe levels (group A) and nine with high (> 25 μmol/l) serum Fe levels (group B) were included. All the subjects were advised to eat a low-Fe diet. The daily Fe intake was reduced approximately 50 % by consumption of the rice-based diet. Haemoglobin, serum Fe, transferrin saturation index (TSI), ferritin and alanine transaminase (EC 2.6.1.2) levels were studied at 1 and 4 months. Dietary Fe intake and body weight were closely monitored. All patients complied with the dietary regimen and at 4 months significant (P < 0·001) reductions from baseline were seen in serum Fe (20 (SD 3) V. 12 (sd 4) μmol/l group A; 30 (sd 3) v. 19 (sd 7) μmol/l group B) and TSI (38 (sd 8) v. 23 (sd 9) % group A; 53 (sd 15) v. 34 (sd 13) %, group B) in both the groups, albeit earlier in group B subjects. Serum ferritin levels, however, reduced only in group A (112 (sd 62) v. 43 (sd 25) ng/ml, P < 0·05) and not in group B. Non-significant reductions in haemoglobin levels were seen in both groups. Alanine transaminase levels reduced significantly (P < 0·05) in both the groups (95 (sd 49) v. 44 (sd 25) IU/l, group A; 82 (sd 16) v. 51 (sd 14) IU/l group B). Thus, a low-Fe diet results in significant reductions in serum Fe and TSI levels, irrespective of baseline Fe levels. This diet should be evaluated to improve the efficacy of interferon therapy in patients with hepatitis B- and C-related chronic liver disease.

The aims of the present study were first to compare the amino acid dilution method performed using labelled animals with that using labelled diets, and second to determine real digestibilities and total ileal endogenous losses of N and amino acids. Two diets containing pea cultivars (Solara and Amino) and a protein-free diet were compared in a 3 × 3 Latin-square experiment. The three pigs were each prepared with an ileo-rectal anastomosis and were continuously infused with [1-13C]leucine. For each cultivar, 15N-labelled and unlabelled diets were formulated. The real digestibility and endogenous losses of leucine were higher when obtained by labelling the pig than by labelling the foodstuff. This was due either to the inadequate estimation of the endogenous protein enrichment in the first case or to the importance of dietary N recycling in the second case. However, in both cases the ileal endogenous losses of N and amino acids were higher than the basal losses determined with the protein-free diet. There were significant differences between the two pea cultivars in terms of phenylalanine and leucine when measured with labelled diets. It is suggested that, although ileal endogenous losses may be underestimated, using labelled feedstuffs is of great interest due to the direct estimation of the individual amounts of amino acids.

Plasma triacylglycerol concentrations increase after the acute ingestion of alcohol (specifically ethanol). However, the effect of ethanol when consumed with a mixed meal has not been well studied. The objective of the present study was to determine the perturbations of lipid metabolism that occur after ingestion of ethanol in combination with a mixed meal of specific fatty acid composition. Blood samples were taken from seven healthy male subjects before and after a mixed meal, with and without ethanol. The specific fatty acid composition of the test meal allowed the fatty acids to be traced into the plasma non-esterified fatty acid pool during the postprandial period. Statistical analysis by repeated measures ANOVA showed significant effects of ethanol. For example, postprandial lipaemia was enhanced after the ethanol test meal compared with the control (P < 0·05), mainly due to increases in triacylglycerol-rich lipoproteins in the flotation range Sf 60–400 (VLDL1) (P < 0·05); those in the range Sf 20–60 (VLDL2) and also Sf > 400 (chylomicrons) were not significantly affected. The later postprandial increase in plasma non-esterified fatty acid concentrations was reduced after the ingestion of ethanol (P < 0·001), but the proportions of palmitoleic acid (a marker of fatty acid content of the test meal) and of linoleic acid (a marker of endogenous lipolysis) were not affected. The results suggest a primary effect of ethanol on the stimulation of secretion of large VLDL particles, which then compete for clearance with chylomicrons by lipoprotein lipase. The results do not support an effect of ethanol on the release of non-esterified fatty acid into the plasma. The suppression of plasma non-esterified fatty acid concentrations during the postprandial period may contribute towards the beneficial effects of moderate ethanol consumption.

Knowledge of levels of fluoride ingestion and excretion is important in planning optimum fluoride therapy for young children. In previous literature, it has been assumed that only about one-third of ingested fluoride is excreted in young children. The aims of the present study were (a) to measure total fluoride intake, urinary fluoride excretion and fluoride balance, and (b) to investigate the effect of air temperature on fluoride intake and urinary fluoride excretion, in young children. Children (4 years old) living in a city, a small town and rural areas of Fars province, Iran, where drinking water contained 0·30–0·39 mg F/l, were invited to participate. Selection of subjects was by random sampling of kindergartens or health centres. The children were surveyed twice, once in summer and once in winter. Diet was obtained by 3 d diaries with interview. Samples of most foods and drinks were analysed for fluoride content. Ingestion of fluoride from toothpaste was estimated for each child. Each child's urine was collected over 24 h and analysed for fluoride content. Seventy-eight of the 116 volunteers completed all aspects of the study, which was conducted in 1995–6. For all children, the mean fluoride ingestion from diet was 0·390 (sd 0·122) mg/d or 0·028 (sd 0·008) mg/kg body weight per d. Fluoride ingestion from diet was higher in summer and higher in rural areas. The mean ingestion of fluoride from all sources was 0·426 (sd 0·126) mg/d and the mean fluoride urinary excretion was 0·339 (sd 0·100) mg/d. The difference between ingestion and urinary excretion was +0·087 (sd 0·143) mg, equivalent to 80 % excretion. Faecal excretion was not estimated. The results indicate fluoride retention at 4 years to be much lower than previously assumed.

The prevalence of obesity is reaching epidemic proportions in many industrialized countries. There is growing evidence that, even if the trigger of this epidemic is found in changes in the environment, genes are interacting with the environment to cause weight gain. Studies of twins reared apart indicate that approximately two-thirds of the variability in BMI is attributed to genetic factors. From prospective studies in Pima Indians we can ascribe 12 % of the variability in BMI to metabolic rate, 5 % to fat oxidation, and another probable 10 % to the level of spontaneous physical activity. These data indicate that at least 40 % of the variability in BMI is related to genetic factors involved in the regulation of food intake and/or volitional activity. This indicates that the most likely successful therapy for obesity may target pathways of the regulation of food intake. Similarly, an environment favouring engagement in physical activity should be promoted.

The in vivo glycaemic responses to six cereal-based foods traditionally consumed in South India were evaluated in patients with non-insulin-dependent diabetes mellitus (NIDDM) and healthy volunteers. All foods contained 50 g carbohydrate and were compared with a 50 g glucose load. Also studied were the in vitro starch digestibility and nutrient composition of the foods. The postprandial responses to the foods at 30, 60 and 120 min were significantly (P < 0·05) lower than those to the reference glucose, in both groups. The peak glucose responses for three foods, i.e. chapatti, idli and poori, occurred 60 min postprandially in both groups. The glycaemic index (GI) values ranged from 67 to 90 in NIDDM and from 44 to 69 in healthy subjects with no significant differences within the groups. Significant relationships were observed between peak responses and area under the curve for foods in patients with NIDDM and in vitro rate of starch hydrolysis (r 0·83, r 0·85, P < 0·05). The GI values predicted using in vitro data were found to be similar to the GI values observed in patients with NIDDM. The GI concept is useful for identifying foods in the habitual Indian diet with attributes of the desired glycaemic effect such as delayed peak rise and low area under the curve.

Vegetarians have a lower incidence of many chronic diseases than omnivores. However, vegetarian diets could potentially result in lower intakes of some minerals, particularly Zn. In a cross-sectional study, dietary Zn intake was measured using 12 d weighed records in ninety-nine vegetarians (ten vegans) aged 18–50 years and forty-nine age- and sex-matched omnivores. In men, the mean daily Zn intake and Zn density values were similar in omnivores, ovolactovegetarians and vegans, but in women they were significantly lower in vegetarians (mean intake 6·8 mg v. 8·4 mg in omnivores) and few achieved the recommended intake. Significantly more vegetarian than omnivorous women had a daily Zn intake < 6 mg (44 % v. 13 %). Mean serum Zn concentrations were similar in female omnivores and vegetarians, despite the differences in intake. However, omnivorous men had a lower mean serum Zn concentration (0·85 μg/ml v. 0·95 μg/ml) and more subjects had levels below the reference range of 0·72–1·44 μg/ml than ovolactovegetarians (P < 0·01). Overall more women than men had low Zn concentrations; and these women generally had intakes below 6 mg/d. There was a significant correlation between serum Zn concentration and dietary Zn density in vegetarians, especially females (P < 0·001), but not in omnivores. Ovolactovegetarians did not have a significantly greater risk of low Zn status than omnivores.

The aim of this present study was to determine the reliability of a buffet-type meal as a measure of spontaneous energy and macronutrient intake. In addition, we evaluated the short-term effects of diet on the composition of the substrate mix oxidized postprandially. Fourteen male subjects had ad libitum access to a variety of foods from a buffet-type meal offered in the laboratory during two identical sessions. The foods comprising the test meal had varying amounts of protein, lipid and carbohydrate. The results showed that there were significant intraclass correlations (ri) for energy (ri 0·97, P = 0·0001), lipid (ri 0·97, P = 0·0001), carbohydrate (ri 0·92, P = 0·0003) and protein (ri 0·82, P = 0·0072) intake between the two meal sessions. Hunger and fullness levels measured immediately before and during 4 h after the meal were identical under the two conditions. In addition, there was no significant difference between the two sessions for RQ and resting energy expenditure, which showed significant reproducibility for measurements obtained immediately before, immediately after, as well as 30 min after, the buffet. This present study demonstrates the high reproducibility of energy and macronutrient intake and oxidation rate values obtained with a buffet-type meal in healthy male subjects and suggests that the use of this test is a reliable method for assessment of macronutrient preferences in the laboratory.

This present study investigated the effect of a 17-week intervention programme with nutrient-dense foods (enriched with vitamins and minerals at 25–100 % of the Dutch recommended dietary allowance) and/or physical exercise in 159 frail elderly subjects (forty-six men, 113 women, mean age 78·7 (SD 5·6) years). Subjects were randomized into four groups: (1) control, (2) nutrition intervention, (3) exercise or (4) both nutrition intervention and exercise. Main outcome variables were sensory perception (smell test and questionnaire), appetite (questionnaire), energy intake (3 d food record) and body weight (on a weighing scale and with dual energy X-ray absorptiometry measurements). At baseline, moderate but significant correlations were found between appetite and energy intake (r 0·30, P < 0·0001), between smell test and smell perception assessed by questionnaire (r 0·40, P < 0·0001) and between lean body weight and energy intake (r 0·50, P < 0·0001). Results after 17-weeks intervention revealed neither change in smell test scores (P = 0·19) nor in appetite (P = 0·17). A slight positive effect of exercise compared with non-exercising groups on energy intake (difference 0·5 MJ, P = 0·05) was shown next to a preserving effect of exercise on lean body mass (+0·08 kg) compared with a decrease (-0·4 kg) in non-exercisers (P < 0·02). The correlation between the change in lean body mass and change in energy intake was 0·18 (P = 0·05). In conclusion, an interesting preserving effect on lean body mass in frail elderly subjects due to 17 weeks of exercise was shown. Since a decline in lean body mass was observed in the non-exercisers, effects may be attributable to change in activity pattern. Changes in lean mass were also slightly, but significantly, correlated with changes in energy intake. In turn, energy intake was not related to a change in reported appetite or sensory perception. Nutrient-dense foods were not able to improve any of the outcome variables in this study.

Trehalose is a disaccharide, the main dietary source being mushrooms. It has been approved as an additive in the preparation of dried food. Isolated intestinal trehalase deficiency is found in 8 % of Greenlanders, but is rare elsewhere. The normal range of trehalase activity and the incidence of isolated trehalase deficiency in the UK have not been reported. Patients (n 400) were investigated for suspected malabsorption. Endoscopic distal duodenal biopsies were taken for histological assessment and maltase, sucrase, lactase and trehalase estimation. Disaccharidase activities were determined by Dahlqvist's technique (). Most patients (n 369) had normal duodenal histology. In these, square root transformation of trehalase activity produced a normal distribution. The normal range (mean ± 2 SD) was 4·79–37·12 U/g protein. One patient had an isolated borderline trehalase deficiency. The thirty-one patients with villous atrophy had significantly reduced disaccharidase activities. With ingestion of a gluten-free diet, maltase, sucrase and trehalase activities recovered to normal in most patients, whereas lactase activity did not. The normal range and very low incidence of isolated enzyme deficiency is comparable with that described in populations from the USA and mainland Europe. Activity is significantly reduced in untreated coeliac disease and recovers with treatment with a gluten-free diet. There is no place for routine determination of trehalase activity in the UK population and there should be no concern over the introduction of trehalose-containing dried foods.

Population-based research on diet, obesity and the metabolic syndrome is faced with accumulating evidence of biases that may profoundly affect results. One potential source of bias, which is often neglected in nutritional epidemiology, arises from self-selected study populations. Subjects who agree to participate in surveys may be at less risk of metabolic syndrome than those who refuse. Analogous to observations in adult populations, studies of schoolchildren have also yielded clear evidence of self-selection. Whether such selection patterns influence analytical results depends on how the biases relate to the dependent and independent variables being studied. Systematic dietary reporting error is another source of bias in studies of nutritional risk factors for disease. While obesity-related under-reporting bias is now well documented, less is known about whether specific foods and nutrients are disproportionately affected. However, two studies employing biomarkers for protein have suggested that obese subjects under-reported the proportion of energy from fat plus carbohydrate. This should alert epidemiologists to the possibility that a dual reporting bias may be present in studies of diet and disease: general under-reporting among obese subjects compounded by food-specific errors. In summary, biases due to self-selection and selective dietary under-reporting may produce consequences in epidemiological studies that are both unpredictable and complex. We conclude this review with recent findings involving dietary fat intake and regional adiposity in a population-based study of women. These preliminary results may have etiological relevance to the development of metabolic syndrome, but multiple biases of the type described previously may also be operating.

The aim of the present study was to improve the reporting of food intake by confronting subjects with their way of reporting food intake, e.g. under-recording and/or undereating. To minimize portion size errors, eighteen female dietitians were recruited as subjects. Energy- and water intake were measured for 1 week with a weighed dietary record. Resting metabolic rate was measured with an open-circuit ventilated-hood indirect calorimeter, and physical activity was measured with an accelerometer for movement registration. Water loss was estimated with 2H-labelled water. Energy balance was checked for by measuring empty body-weight 1 week before the start, at the start and at the end of the recording week. In the first part of the study, the change in body weight in the non-recording week was 0·14 kg and in the recording week -0·45 kg (P 0·02), indicating 12 % undereating. Total water intake closely matched measured water loss, indicating a high recording precision. There was under-reporting of habitual food intake that could be fully explained by undereating. In the second part of the study, subjects were confronted with these results and the protocol was repeated. This time there was no significant change in body weight in the recording week, indicating no undereating. The reporting of habitual food intake had been improved. In conclusion, in the studied group of highly motivated lean women, under-reporting of habitual food intake (here due to undereating) could be eliminated by confrontation with the results of this phenomenon.

An evaluation of general models that describe gas production profiles is presented. The models are derived from first principles by considering a simple three-pool scheme and permit the extent of ruminal degradation to be calculated, as described in the companion paper. The models evaluated were the generalized Mitscherlich, simple Mitscherlich, generalized Michaelis–Menten, simple Michaelis–Menten, Gompertz, and logistic. Five sets of gas production data consisting of 216 curves, obtained using a wide range of feeds (including straw, hay, silage, grain and various byproducts), were analysed to study the performance of these gas production models. Application of the non-sigmoidal models (simple Mitscherlich and Michaelis–Menten) to the data resulted in convergence problems and these models were found to be inadequate in many cases. Based on results of a pairwise comparison between models (variance ratio test), ranking of residual mean squares, lack-of-fit test, and of analyses of residuals, the generalized Mitscherlich and the generalized Michaelis–Menten models seemed particularly suited because of their flexibility to encompass sigmoidal and non-sigmoidal shapes of gas production profiles, whether symmetrical or not.

Equations to describe gas production profiles, obtained using manual or automated systems for in vitro fermentation of ruminant feeds, were derived from first principles by considering a simple three-pool scheme. The pools represented were the potentially degradable and undegradable feed fractions, and accumulated gases. The equations derived and investigated mathematically were the generalized Mitscherlich, generalized Michaelis–Menten, Gompertz, and logistic. They were obtained by allowing the fractional rate of degradation to vary with time. The equations permit the extent of ruminal degradation (hence the supply of microbial protein to the duodenum) to be evaluated, thus linking the gas production technique to animal production.

Anaerobic batch culture fermenters were used for a preliminary screening of the in vitro utilization by human gut microflora of dextran and novel oligodextrans (I, II and III) produced in the University of Reading (UK). Glucose and fructooligosaccharides (FOS) were used as reference carbohydrates. As expected, FOS acted as a good prebiotic in that it selectively increased numbers of bifidobacteria in the early stages of the fermentation. Dextran and oligodextrans each resulted in an enrichment of bifidobacteria in the batch cultures, with high levels of persistence up to 48 h. They also produced elevated levels of butyrate ranging from 5 to 14·85 mmol/l. To more effectively simulate conditions that prevail in different regions of the large intestine, a three-stage continuous culture cascade system was used to study further the fermentation of dextran, a low-molecular-mass oligodextran (IV) and maltodextrin. Oligodextran IV was shown to be the best substrate for bifidobacteria and lactobacilli with steady-state populations of bifidobacteria and lactobacilli being higher in all three vessels of the gut model than the respective populations resulting from dextran and maltodextrin. A maximum difference of 1·9 log was observed in vessel 1 for both bifidobacteria and lactobacilli in the case of dextran fermentation, while 1·4 log and 0·8 log in vessel 3 were the maximum differences for bifidobacteria and lactobacilli when maltodextrin was used as the carbohydrate source. Moreover, dextran and oligodextran appeared to stimulate butyrate production, with a maximum production up to 25·39 mmol/l in vessel 3 when fermenting dextran, followed by 21·70 mmol/l in the case of oligodextran IV and only 12·64 mmol/l in the case of maltodextrin.

The energy cost of locomotion of four Iberian pigs was measured in two experiments conducted when the animals averaged 41·3 (se 0·1) kg (first experiment) and 84·1 (se 0·1) kg (second experiment). The heat production of the pigs was determined when standing or walking at a speed of 0·555 m/s on a treadmill enclosed in a confinement-type respiration chamber, on different slopes (-10·5, 0, and +10·5 % in the first experiment, and -5·25, 0 and +10·5 % in the second experiment). The energy costs of locomotion, estimated from the coefficients of linear regressions of heat production per kg body weight (BW) on distance travelled, were in the first experiment 2·99, 3·31 and 5·88 J/kg BW per m for -10·5, 0, and +10·5 % inclines respectively, and 2·56, 2·84 and 7·13 J/kg BW per m for -5·25, 0 and +10·5 % inclines respectively, in the second experiment. The net energy cost of locomotion on the level appeared to be independent of live weight, attaining a value of 2·98 J/kg BW per m. Also, it was found that within experiments the net energy cost of walking on negative slopes was similar to that for locomotion on the level, indicating that no energy was recovered on vertical descent. Mean values were 3·11 and 2·72 kJ/kg BW per m for the light and heavy pigs respectively. The energy cost of raising 1 kg BW one vertical metre was found to be 27·1 J/kg BW per m in the first experiment and 40·0 J/kg BW per m in the second experiment. Correspondingly, the calculated efficiency for upslope locomotion appeared to decline with increasing BW, resulting in average values of 36·2 and 24·5 %.

The role of high-fat diets in weight gain and obesity has been questioned because of inconsistent reports in the literature concerning the efficacy of ad libitum low-fat diets to reduce body weight. We conducted a meta-analysis of weight loss occurring on ad libitum low-fat diets in intervention trials, and analysed the relationship between initial body weight and weight loss. We selected controlled trials lasting more than 2 months comparing ad libitum low-fat diets with a control group consuming their habitual diet or a medium-fat diet ad libitum published from 1966 to 1998. Data were included from 16 trials with a duration of 2–12 months, involving 1728 individuals. No trials on obese subjects fulfilled the inclusion criteria. The weighted difference in weight loss between intervention and control groups was 2.55 kg (95 % CI, 1.5–3.5; P < 0.0001). Weight loss was positively and independently related to pre-treatment body weight (r = 0.52,P < 0.05) and to reduction in the percentage of energy as fat (0.37 kg / %, P < 0.005) in unweighted analysis. Extrapolated to a BMI of about 30 kg/m2 and assuming a 10 % reduction in dietary fat, the predicted weight loss would be 4.4 kg (95 % CI, 2.0 to -6.8 kg). Because weight loss was not the primary aim in 12 of the 16 studies, it is unlikely that voluntary energy restriction contributed to the weight loss. Although there is no evidence that a high intake of simple sugars contributes to passive overconsumption, carbohydrate foods with a low glycaemic index may be more satiating and exert more beneficial effects on insulin resistance and cardiovascular risk factors. Moreover, an increase in protein content up to 25 % of total energy may also contribute to reducing total energy intake. In conclusion, a low-fat diet, high in protein and fibre-rich carbohydrates, mainly from different vegetables, fruits and whole grains, is highly satiating for fewer calories than fatty foods. This diet composition provides good sources of vitamins, minerals, trace elements and fibre, and may have the most beneficial effect on blood lipids and blood-pressure levels. A reduction in dietary fat without restriction of total energy intake prevents weight gain in subjects of normal weight and produces a weight loss in overweight subjects, which is highly relevant for public health.

Increased concentrations of n-3 polyunsaturated fatty acids (PUFA), namely eicosapentaenoic acid (20 : 5; EPA) and docosahexaenoic acid (22 : 6; DHA), have been shown to be beneficial in coronary artery disease (CAD). In the present study, the relationships between fish intake and concentrations of serum EPA and DHA and the effects of these fatty acids on serum lipids and lipoproteins were investigated. Two groups of men, one living in a fishing village and the other in a farming village, participated in this study. The daily fish consumption was ten times greater in the fishing village group than in the rural village group and the mortality from IHD in the rural village was four times higher. Serum concentrations of EPA and DHA were significantly higher in the fishing village group (P < 0·001). In this group, the serum concentration of arachidonic acid (20 : 4; AA), was significantly lower (P < 0·001), and the ratio EPA : AA was twice that of the rural village (P < 0·001). Moreover, in the fishing village group, the serum triacylglycerol and total cholesterol levels were significantly lower than those observed in the rural village (P < 0·01 and P < 0·05 respectively). In the fishing village group the serum LDL-cholesterol concentration was also lower, although the difference was not significant. Our results reinforce the hypothesis that a high intake of n-3 PUFA provides protection against CAD.