Objective This study tests the hypothesis that natural selection is associated with type 2 diabetes (T2D)‐associated mortality and fertility in a rural isolated Zapotec community in the Valley of Oaxaca, southern Mexico. Participants and Methods Mortality data and related demographic and genealogic information were linked with data for fertility, prereproductive mortality and family history of mortality attributed to T2D. Physician verified T2D mortality (n = 27) between 1980 and 2009 and imputed T2D (n = 70) from cardiovascular mortality (68% random sample) and renal failure (44% random sample). Bootstrapping was used to obtain a robust variance estimate in survival analysis and multivariate analysis of variance. Results Estimated maximum natural selection by Crow's Index occurred circa 1930 and was relaxed after this time in the study population. Cox‐regression survival analysis of T2D mortality with covariates (family history of T2D, cardiovascular disease, renal failure) indicated a significant hazard ratio (HR = 5.95, 95% CI: 1.38–25.67, p < .008) for the increase in T2D in 2000 to 2009. Survival analysis of imputed T2D resulted in a significant HR of 2.03 (95% CI: 1.08–3.85, p = .01) for the increase in T2D in the 2000 to 2009 cohort (reference group: 1980–1989). Mean number of live born offspring was lower among T2D (n = 27, 4.04 ± 3.85 SD) compared to non‐T2D (n = 199, 5.30 ± 3.48) groups (p < .08). Mean number of live born offspring was lower (p = .003) among imputed T2D compared to non‐T2D groups (4.10 ± 3.44 vs. 5.62 + 3.50). Discussion T2D‐associated mortality increased in frequency as natural selection decreased, and favored offspring survival of non‐T2D descedants. The results indicated statistically significant directional selection against T2D and imputed T2D to this population isolate.