Evidence is accumulating that low androgen concentrations are a cause of rheumatoid arthritis. This would explain a number of established features of the epidemiology of the disease. These include: (a) the variation of disease activity with pregnancy; (b) the variation of age at onset by sex; (c) the variation by sex with HLA-B15; (d) the association with bone mineral density; and (e) the differing time trends in incidence rates by sex. It is argued, moreover, that if one makes a plausible assumption--namely, that women who choose oral contraceptives have high androgen concentrations at the time they first make this choice--then an explanation becomes available for the confusion about the relation between rheumatoid arthritis and oral contraception. Grounds are adduced for that assumption. If this line of reasoning is substantially correct it also has implications for the relations between rheumatoid arthritis and smoking and consumption of alcohol.