This study was performed to review reports of the descriptive epidemiology of chronic arthritis in childhood and to analyze the factors that may explain differences in its reported frequency. Articles were retrieved by searching MEDLINE and EMBASE under the following index terms: juvenile rheumatoid arthritis (JRA), juvenile chronic arthritis (JCA), spondyloarthropathy, epidemiology, prevalence, and incidence. For reports published between 1977 to 1982, the Index Medicus was used. All original articles that provided prevalence or incidence rates, population size, or number of cases, were reviewed and entered into the analysis. Variables analyzed were disease prevalence and incidence. Modifier variables investigated were diagnostic criteria, source population, geographic origin of the report (Europe or North America), duration of the study, and race of the population studied. Diagnostic criteria had no effect on reported prevalence or incidence rates. Prevalence per 100,000 at risk obtained from population studies (132, 95% CI: 119, 145) was significantly higher than values derived from practitioner- (26, 95% CI: 23, 29) or clinic-based studies (12, 95% CI: 10, 15) (P = .02). North American clinic-based studies had higher prevalence values compared with European reports (32, 95% CI: 26, 38 versus 8, 95% CI: 5, 11, P = .009). None of the factors analyzed accounted for the variability in reported incidence rates. An effect of race was detected only in the distribution of patients among onset subsets. Thus, the percentage of patients with pauciarticular JRA was highest in series of North American and European caucasian patients (58, 95% CI: 56, 60) compared with series of East Indian (25, 95% CI: 20, 31), native North American Indian (26, 95% CI: 15, 37), or other races (31, 95% CI: 28, 35) (P = .001). In contrast, the percentage of patients with polyarticular JRA was lowest in the former (27, 95% CI: 25, 28) compared with the other racial groups (East Indian, 61, 95% CI: 55, 66; native North American Indian, 64, 95% CI: 53, 76; other races, 34, 95% CI: 30, 38) (P = .004). Although an effect of source population on reported prevalence was confirmed, the effect of geographic origin suggests that environmental or ethnic differences also may influence the prevalence of chronic arthritis in children. Differences in the percentages of patients with pauciarticular and polyarticular JRA may reflect racial differences in the prevalence of these conditions.