Objective: To determine whether a reported positive association between parity and the development of non-insulin-dependent diabetes mellitus (NIDDM) and impaired glucose tolerance (IGT) is reproducible in other populations. Research Design and Methods: We investigated the relationship in data from population-based surveys in four Pacific and Indian Ocean island nations. Women ≥ 40 yr of age at the time of the survey, excluding those in whom diabetes developed before 40 yr of age, were included in this study of Micronesians from Nauru (n = 204) and Kiribati (n = 562), Fiji Melanesians (n = 390), Fiji Indians (n = 247), and mixed-ethnic Mauritians (n = 1333). Subjects in each survey underwent a 75-g oral glucose tolerance test, and glucose tolerance status was ascertained with 1985 World Health Organization criteria. Obstetric information and family history of diabetes were determined by interview. Results: Age and body mass index (BMI)-adjusted mean parity increased slightly with worsening glucose tolerance in only two groups, decreased in one group, and was inconsistent in the other two (none were statistically significant). We also found an inconsistent relationship between the number of full-term pregnancies and the prevalence of IGT and NIDDM, although in each population, there was a higher prevalence of NIDDM in the highest parity group (≥ 10 pregnancies) compared with the lowest parity group (1-3 pregnancies). In logistic regression analyses accounting for age, BMI, and family history of diabetes, odds ratio estimates for NIDDM and IGT associated with each pregnancy were not significantly greater than unity. Conclusions: The results indicate that there is little if any independent association between parity and the development of abnormal glucose tolerance in these populations.