The overall objective of a series of experiments to investigate ‘metabolic stress’ was to examine the relationships between ‘metabolic load’, disease and other parameters associated with the welfare of the dairy cow. In the main, these used several well controlled herd based studies complimented with more basic and strategic investigations. In this paper we compare and contrast practical aspects of health and welfare in two high genetic merit herds managed at the extremes of inputs and outputs for dairy farming in south-west Scotland. The hypothesis was that high output herds would have more health and welfare problems than low input herds. Two herds (70 Holstein-Friesian cows each) at SAC Acrehead Dumfries of a similar genetic background (overall in the top 5% of UK cows by PIN and ITEM), were housed in identical buildings and tended by the same herdsman. Both herds had autumn- and spring-calving cattle. The ‘low input’ herd (LI) was given a minimum of concentrate (approx. 0.5 t per cow per year) and milked twice a day and had a restricted quota of 385 000 l. The ‘high output’ herd (HO) was managed for high yields (unrestricted quota) and was given concentrates (2 t per cow per year) and forage ad libitum and milked three times daily. In 1995-96 the sole source of winter forage was grass/clover silage (LI) or grass silage (HO) but in 1996-1998 ensiled cereal and fodder beet were included in both diets. ‘Metabolic load’ could only be inferred from overall inputs, milk outputs, weight loss, body condition score and behaviour. There were significant differences in 305-day lactation yields between herds, and season of calving especially in 1995-96 (LI autumn; 5952 l at 30 g/kg protein (P); LI spring; 5741 l, 32.5 g/kg P; HO autumn; 9541 l at 32.8 g/kg P; HO spring; 8402 l, 32.6 g/kg P). LI weight and body condition-score losses were greatest in this year and behavioural studies showed substantial differences in feeding time (HO < LI, P < 0.05) and total lying time (LI < HO; P < 0.05). However these differences were much less marked in subsequent years. There was a significant difference in the prevalence and incidence of clinical lameness between herds (HO > LI; P < 0.05) and season (autumn > spring P < 0.05) but not for mastitis or metabolic disease. An in-depth study of subclinical claw horn lesion development in first calving heifers showed significant differences between herds in 1996-97 (LI > HO, P < 0.05) but none in 1995-96. There was a significant difference for season in both years (autumn > spring, P < 0.05). Analysis of blood biochemistry parameters of samples taken at approximately 1 month after calving showed some significant differences between LI and HO generally indicating a greater ‘metabolic load’ for LI. Although the full effects of ‘metabolic load’ on immune function and reproduction are dealt with elsewhere our preliminary data showed no significant differences between herds for the former but some significant differences for the latter, in particular there were differences in aspects of the progesterone profiles between herds and more importantly between seasons. However these latter differences were not clearly reflected in conception rates. It was concluded that the hypothesis was not fully sustained and that both systems had pitfalls in terms of welfare. The three major areas causing difficulties for both systems were the need first to ensure adequate intake of forage; secondly to limit the environmental challenge to the feet and udder and finally to marry these systems to the factors limiting reproduction, primarily calving season and ability of reproduction management.