1. Deficits in weight- and length-for-age, and serum albumin and transferrin concentrations were determined for children who were either marginally undernourished (twenty-five children) or suffering from either marasmus (thirty-two children) or kwashiorkor (twenty-six children) defined according to the Wellcome Classification (Waterlow, 1972). The measurements were also made in eight children with kwashiorkor after the loss of oedema, and in sixteen children who were recovering from either marasmus or kwashiorkor.
2. The mean concentration of serum albumin was similar for children from the ‘under-nourished’ group and from the group with marasmus, but was significantly reduced in those with kwashiorkor.
3. The concentration of serum transferrin was significantly reduced in both the group of children with marasmus and those with kwashiorkor. The serum transferrin concentration was significantly lower in children with kwashiorkor when compared with the level in those with marasmus.
4. Seventeen children (seven with kwashiorkor and ten with marasmus) died. These children were neither lighter nor shorter than the severely malnourished children who survived. The concentration of serum albumin was not lower in the children who died than in those who survived.
5. In contrast to the results for serum albumin concentrations, the children who died had significantly lower levels of serum transferrin than those who survived.
6. There was a significant linear relationship between serum transferrin concentrations and the deficits in length-for-age (P < 0·05) and weight-for-length (P < 0·001) in the marginally undernourished children. The deficit in weight-for-length was also linearly related to the serum transferrin concentrations (P < 0·001) in children recovering from severe malnutrition.
7. It is suggested that the measurement of serum transferrin concentrations provides an index of severity in severely malnourished children, and should prove useful in field assessments of nutritional status.