A longitudinal study of four birth cohorts (1935–1944, n = 1,598; 1945–1954, n = 2,208; 1955–1964, n = 2,781; and 1965–1974, n = 2,230) found that succeeding generations had a higher prevalence of arthritis. Various risk factors were associated with arthritis but did not account for this cohort effect. The benefits of improved education, increased income and smoking cessation on arthritis prevalence partially overcame the detrimental effects of increasing BMI. Further analysis suggested that obese individuals had an earlier age of arthritis onset compared with individuals of normal weight.