The most destructive fungal disease of soybean [Glycine max (L.) Merr.] is Phytophthora root rot. The most common method of protecting soybean from Phytophthora root rot utilizes specific resistance genes, but continuous selection pressure exerted on the pathogen frequently causes resistance genes in soybean to become ineffective. Infected plants which exhibit tolerance often yield nearly m well as uninfected resistant plants and prevent the buildup of races capable of overcoming resistance genes. Experiments were initiated to obtain information concerning the genetics of tolerance to Phytophthora root rot. The objectives were to: (i) compare parent-offspring regression, correlation, and realized heritability estimates of tolerance, (ii) compare heritability and phenotypic variation among single-, three-way, and double-cross populations, and (iii) correlate non-infected yields with tolerance and maturity. Ten [F.sub.3:4] populations were screened for tolerance in the [F.sub.3] and [F.sub.4] generations by the inoculum layer method. Heritability estimates were low, ranging from 0.17 to -0.10 when obtained through parent-offspring regression. These estimates were not separable from zero because of the magnitude of their standard errors. Parent-offspring correlation estimates were larger than parent-offspring regression estimates, and ranged from 0.43 to -0.19. Realized heritability was estimated by selecting the 20% highest and lowest tolerant lines within each population. These results were generally larger, ranging from 0.80 to -0.13, and suggested that relatively high tolerant lines may be identified through selection. The number of parents composing a population did not influence heritability or phenotypic variation within populations. Most correlations among yield, tolerance, and maturity were non-significant. Tolerance levels within populations should not vary with maturity. The inoculum layer method may not be sensitive enough to allow accurate screening of germplasm with low tolerance levels.