Reports of positive associations between allozymic heterozygosity and measures of fitness are routine, but it has not been possible to distinguish between the two preeminent explanations of the phenomenon, dominance and overdominance. We tested several of the assumptions of these hypotheses in our study of the relationship between electrophoretic genotype and three life history traits in loblolly pines (Pinus taeda L.). Traits examined included the survival and growth of selfed and outcrossed progeny of 45 maternal trees, and maternal fecundity, measured as the number of surviving progeny per mother tree. Inbreeding depression was severe; the relative fitness of the selfed progeny was only 8% that of the outcrossed progeny. We found a heterozygote fecundity advantage, which should have resulted in an excess of rare alleles in the progeny. Instead, there was evidence of severe survival selection against rare alleles in both heterozygous and homozygous forms. The deficit of rare alleles averaged 69 and 50% in the selfed and outcrossed progeny, respectively. The one allele in the sample that we should have suspected of being maintained by overdominance (a PGI2 mid-frequency allele) appeared to be overdominant for outcrossed height growth and probably for fecundity as well. Multiple-locus genotype explained very little of the variation in growth, however, and rather than seeing evidence for overdominance as a force in maintaining most of the observed polymorphism, we were left to explain, in the face of the severe survival selection, why the rare alleles were present at all. Projection of the stand into the future through computer simulation showed how balancing selection acting on differential growth, fecundity, and mortality among genotypes could, over the life of the stand, account for the maintenance of the rare alleles in the population.