Cleistogamy is a mating system in which plants produce some proportion of closed, autonomously self-pollinating flowers. Cleistogamous flowers differ from chasmogamous flowers, which are open flowers capable of outcrossing. Both dimorphic cleistogamy (cleistogamous and chasmogamous flowers produced on the same plant) and complete cleistogamy occur. Cleistogamy has been hypothesized to be a bet hedging strategy for reducing risk in the face of unpredictable pollinator availability. However, conflicting results across species and challenges connecting theory to data have prevented researchers from proving that cleistogamy is bet hedging. To test the bet hedging hypothesis, we investigated the distribution of over 400 cleistogamous species through biogeographical analyses. We find that cleistogamy is more prevalent in cooler, more variable environments. Additionally, we find that among cleistogamous species, complete cleistogamy is more likely to occur in warmer, more stable, tropical and subtropical environments. We hypothesize that the difference in distribution between complete and dimorphic cleistogamy may be driven by the opposing forces of selection to increase cleistogamy proportion and extinction risk, which we test using a heuristic Markov transition model. We conclude that the distribution of cleistogamy suggests that the strategy has evolved in variable environments, consistent with expectations for bet hedging.