Dimorphic cleistogamy is a specialized form of mixed mating system where a single plant produces both open, potentially outcrossed chasmogamous (CH) and closed, obligately self-pollinated cleistogamous (CL) flowers. Typically, CH flowers and seeds are bigger and energetically more costly than those of CL. Although the effects of inbreeding and floral dimorphism are critical to understanding the evolution and maintenance of cleistogamy, these effects have been repeatedly confounded. In an attempt to separate these effects, we compared the performance of progeny derived from the two floral morphs while controlling for the source of pollen. That is, flower type and pollen source effects were assessed by comparing the performance of progeny derived from selfed CH vs. CL and outcrossed CH vs. selfed CH flowers, respectively. The experiment was carried out with the herb Ruellia nudiflora under two contrasting light environments. Outcrossed progeny generally performed better than selfed progeny. However, inbreeding depression ranges from low (1%) to moderate (36%), with the greatest value detected under shaded conditions when cumulative fitness was used. Although flower type generally had less of an effect on progeny performance than pollen source did, the progeny derived from selfed CH flowers largely outperformed the progeny from CL flowers, but only under shaded conditions and when cumulative fitness was taken into account. On the other hand, the source of pollen and flower type influenced seed predation, with selfed CH progeny the most heavily attacked by predators. Therefore, the effects of pollen source and flower type are environment-dependant and seed predators may increase the genetic differences between progeny derived from CH and CL flowers. Inbreeding depression alone cannot account for the maintenance of a mixed mating system in R. nudiflora and other unidentified mechanisms must thus be involved.