There are many reports of dioecious flowering plants with positive spatial association of gender, such that males and females tend to occur at different sites. This pattern has been interpreted as demonstrating environmenal heterogeneity, the two sexes having different habitat preferences. However, it is not clear: (a) to what extent the literature reflects the greater likelihood of publishing striking positive results; and (b) whether the association of stems of like gender within habitats is caused by differences in microhabitat preference rather than by vegetative ramification. We surveyed natural populations of ten sexually heteromorphic species in southern Quebec, and found that five showed positive association and four showed no association between the genders. With one exception (Silene cucubalus) these results appeared to reflect the presence or absence of vegetative ramification. We conclude that the demonstration of spatial association cannot be used to infer the existence of microenvironmental heterogeneity except in species with no capacity for vegetative ramification. The tenth species, Silene alba, showed negative association, with an excess of neighbouring pairs of unlike gender, but we were unable to reproduce this phenomenon experimentally.