Quantitative and spatial data for orchid pollination are scarce and may be important tools for reintroduction and conservation; however, conclusions cannot be drawn on the basis of the typically infrequent and unpredictable pollination events. We carried out a novel, retrospective, spatial analysis of the pollination of the entire population of two miniature orchids, Erycina crista-galli and Notylia barkeri, on coffee bushes in plantations at 900 m in Soconusco, south-eastern Mexico. The numbers of mature flowering plants of both species in the experimental site were similar. Notylia barkeri produced nearly four times as many flowers, but a similar proportion of the total number of flowers produced was pollinated (1.23% and 1.48% for N. barkeri and E. crista-galli, respectively). An estimated 29 919 977 (±4 983 995) seeds were produced by N. barkeri, nearly 12 times more than E. crista-galli at 1 009 414 (±147 000). The pollinators of N. barkeri chose flower clusters at random and pollinated various flowers within a patch, whereas the pollinators of E. crista-galli chose patches of flowers slightly more systematically, with less dependence on flower density, and appeared to dedicate less attention to each patch. For both species, pollinators slightly favoured larger clusters of flowers and left many individual and groups of flowers unvisited. To restore populations of these orchids in coffee plantations as a replacement habitat, N. barkeri should be planted in small, separate groups and E. crista-galli in larger groups of individuals, dispersed regularly throughout the selected site to maximize the possibility that the flowers will be discovered by pollinators. © 2008 The Linnean Society of London, Botanical Journal of the Linnean Society, 2008, 158, 448–459.