1. Caffeine has been used to determine acetylator phenotype for some 15 years but the interpretation of metabolic ratios with this substance raises theoretical and methodological issues. 2. N-acetyltransferase type 2 (NAT2) status was assessed in 23 young healthy subjects using both caffeine overnight and spot urine samples, and sulphadimidine. 3. Frequency distribution analysis of the metabolic ratios of NAT2, indicated two distinct groups for sulphadimidine, and for caffeine spot but not overnight samples. Spearman's rank correlation values were low indicating differences between the data sets for sulphadimidine and spot caffeine samples. Correlation between the two urine collection periods for caffeine was poor. 4. The complex metabolism of caffeine may compromise its value as a probe for determining acetylator phenotype until the effects of important variables are better understood.