Indeterminate bean (Phaseolus vulgaris L.) lines derived from crosses between small-seeded indeterminate and large-seeded determinate genotypes were used to estimate the heritabilities and phenotypic correlations for seed yield and seed yield components. The F2 generation of six bean populations was planted at the Fortuna Substation, Juana Díaz, Puerto Rico, in October 1984. Seed yield per plant, 100 seed weight, number of pods per plant, and number of seed per pod were measured for 50 plants selected at random from each population. A total of 50 F3 plant rows of each population were planted in February 1985 at the Fortuna Substation, and in March 1985 on a small farm in the Constanza valley of the Dominican Republic. Narrow sense heritabilities were estimated by using parent-offspring regressions of the F2 and F3 generation, and phenotypic correlations were estimated by using means of the F3 lines. Mean seed yields per plant of the indeterminate F3 lines were significantly less than the indeterminate parents, whereas the 100-seed weights of the indeterminate F3 were significantly less than the determinate parents. Narrow sense heritabilities for seed yield and seed yield components were intemediate to low. These results indicate that selection for greater seed yield would be more effective by evaluating advanced lines in replicated trials. Spearman rank correlations between locations for seed yield and seed yield components varied among populations. Multilocation testing of advanced generation lines may be the most effective way to identify bean genotypes that perform well in contrasting environments.