The use of phenytoin has increased among the rural black population in South Africa, many of whom have albumin concentrations below the accepted reference range of 35-50 g/L, related to a combination of malnutrition and late-presenting renal and hepatic disease. Because albumin concentration has a major effect on the proportion of free phenytoin in the extracellular fluid, we instituted a study of the extent of hypoalbuminemia and of the difference between "total" phenytoin (measured by immunoassay), and "corrected" phenytoin (calculated using the Sheiner-Tozer equation, which is based on a mean albumin of 40 g/L). The differences were significant (higher than 20%) in 37% of patients and led us to propose that in populations in which there is a high proportion of patients who are hypoalbuminemic, it is corrected rather than total phenytoin that should be the value reported.