Various aspects of the reproductive success of Garcinia atroviridis Griff. were studied. Controlled pollination experiments were carried out in an orchard located in Songkhla province, southern Thailand, from February to July 2003. Floral longevity, stigma receptivity, and pollen viability were examined before carrying out the experiments. Three pollination treatments were compared: open pollination, manual pollination with bags, and bags without pollination (apogamy). Although there was no significant difference in the initial fruit set, bagged and manual pollination produced a significantly greater fruit drop rate than apogamy or natural pollination at one week after the flowers had been pollinated. On the other hand, the apogamy treatment had a greater fruit drop rate than natural and manual pollination treatments before fruit maturation. In addition, unpollinated bagged flowers bore fewer and smaller fruit than naturally and manually cross-pollinated flowers. Although the fruits from unpollinated flowers were capable of asexual seed formation, they produced fewer seeds and had poorer seed quality (defined as average fresh weight and germination rate) than those from the other treatments. The occurrence of asexual and sexual reproduction was also studied using Random Amplification of Polymorphic DNA (RAPD) analysis and by comparing the patterns of bands produced from DNA extracted from the offspring of the naturally cross-pollinated fruits. On average, 58% of the offspring had a genetic constitution identical to that of the maternal parent (ranging from 36% to 87%), indicating that some offspring were produced without prior fertilisation. However, the remainder showed polymorphism, demonstrating the occurrence of sexual reproduction. These findings indicate that facultative apomixis occurred in the study population. However, a residual sexuality was important for fruit production, fruit size, normal seed set and seed quality.