In gynodioecious plants, seed offspring from hermaphrodites often perform less well than those from females. This lower performance sometimes can be attributed to inbreeding by hermaphrodites or to relatively greater provisioning of individual seeds by females. However, these hypotheses are not explanatory when only outcrossing occurs and when individual seeds of the two morphs are equally well provisioned. Three hypotheses may explain the lower fitness of seed offspring from hermaphrodites in such cases. The morphology hypothesis states that the opportunity for gametophytic selection is lower within flowers of hermaphrodites compared to flowers on females, because the perfect flowers of hermaphrodites are relatively short-styled. The cytotype hypothesis states that the performance difference is directly caused by an individual's cytotype, whose frequency in the population may differ for the two sex morphs. The pleiotropy hypothesis states that negative pleiotropic effects of nuclear restorer alleles or alleles hitchhiking with them are expressed more often by offspring from hermaphrodites. We performed two experiments using the gynodioecious plant Silene acaulis to contrast these hypotheses. In our first experiment we contrasted the morphology and pleiotropy hypotheses by performing controlled pollinations and subsequently planting seeds in both the greenhouse and field. Hermaphrodites of S. acaulis can produce both pistillate and perfect flowers, which allowed us to determine whether flower morphology affects offspring survivorship independent of the sex of the maternal parent. We found that neither seed mass nor germination differed between seeds from females and hermaphrodites. Offspring from pistillate flowers on hermaphrodites did not differ significantly in their survival compared to offspring from perfect flowers on hermaphrodites, but had lower survivorship compared to offspring from pistillate flowers on females, refuting the morphology hypothesis. In a second experiment, we compared offspring survival of full-sibling pairs of females and hermaphrodites (who shared the same cytoplasm) to contrast the cytotype and pleiotropy hypotheses. We found that seed offspring from females and hermaphrodites that shared the same cytoplasm differed in their survival, which is counter to the prediction of the cytotype hypothesis. In both experiments, the sex of the maternal parent significantly affected offspring survival, with seed offspring from hermaphrodites surviving less well than those from females. These results support the pleiotropy hypothesis. We conclude by discussing alternative ways of thinking about negative pleiotropic effects of nuclear restorers or “the cost of restoration.”