The California drought of 1975-77 has been correlated with unusual size changes in populations of two species of Euphydryas butterflies. Several populations became extinct, some were dramatically reduced, others remained stable and at least one increased. These differences in population dynamic response are not concordant with predictions made earlier that populations with heavy density-dependent mortality should be more stable in the face of drought than unregulated populations. The different responses are related to the fine details of the relationships between the insects and their host plants, relationships which are variable between populations. Revised predictions are given in the light of better knowledge of the variability and complexity of these insect-host relationships. The diversity of responses underlines the dangers of generalizing about "the ecology" of a taxonomic species. The extinctions support the view that such events are frequent and significant in the biology of populations.