Arylamine N-acetyltransferase catalyses the N-acetylation of primary arylamine and hydrazine drugs and chemicals. N-acetylation is subject to a polymorphism and humans can be categorized as either fast or slow acetylators according to their ability to N-acetylate polymorphic substrates in vivo. Previously, slow acetylation has been linked to four distinct polymorphic N-acetyltransferase (pnat) alleles each of which contains one or more point mutations within the coding region of the pnat gene. One new rare slow variant of pnat has been identified by cloning and sequencing the pnat DNA from an individual whose NAT phenotype was determined by in vivo acetylation of the polymorphic substrate sulphamethazine. This allele, designated S1c, differs from the wild type fast allele at nucleotide positions 341 and 803. A second new rare slow allotypic variant, designated S3, has been identified by resistance of the pnat specific DNA to digestion with the restriction enzymes Fok I and Bam HI. A method of genotyping individuals for the arylamine N-acetyltransferase (NAT) polymorphism is presented which correctly predicts the phenotype of greater than 95% (21 of 22) of individuals as measured by the extent of acetylation of sulphamethazine in urine. This refined genotyping method was applied to a clinical population of 48 Caucasians with classical or definite rheumatoid arthritis each receiving daily between 150 and 500 mg of the anti-rheumatic drug, D-penicillamine. There is no difference in the N-acetyltransferase phenotype of the individuals who developed proteinuria and the control group with no adverse effects.