The heritability of age- and length-at-maturity was estimated in two populations of mosquitofish that shared ancestors in 1905. Three techniques were used-estimates of family effects in the first laboratory-reared generation, regression of the offspring produced in the second laboratory-reared generation on their parents, and a full-sib/half-sib cross that produced the second generation. The parent-offspring regressions and the full-sib/half-sib cross gave concordant results. Heritability of length-at-maturity was .79 ± .65 in fish from Reservoir 33, a reservoir that undergoes dramatic fluctuations in water level. Heritabilities of age-at-maturity were low and not significant in either population, and the heritability of length-at-maturity was low and not significant in fish from Twin Reservoir, a reservoir that has maintained stable water levels for 70 years. These results indicate that conclusions drawn from family effects detected in the first laboratory-reared generation were misleading because additive genetic variance was confounded by significant maternal effects and effects of common environments early in life. They also suggest, but do not demonstrate, that selection pressures on length-at-maturity are probably stronger in Twin Reservoir than in Reservoir 33.