Generics are statements such as "tigers are striped" and "ducks lay eggs". They express general, though not universal or exceptionless, claims about kinds (Carlson & Pelletier, 1995). For example, the generic "ducks lay eggs" seems true even though many ducks (e.g. the males) do not lay eggs. The universally quantified version of the statement should be rejected, however: it is incorrect to say ""all" ducks lay eggs", since many ducks do not lay eggs. We found that adults nonetheless often judged such universal statements true, despite knowing that only one gender had the relevant property (Experiment 1). The effect was not due to participants interpreting the universals as quantifying over subkinds, or as applying to only a subset of the kind (e.g. only the females) (Experiment 2), and it persisted even when people judged that male ducks did not lay eggs only moments before (Experiment 3). It also persisted when people were presented with correct alternatives such as "some ducks do not lay eggs" (Experiment 4). Our findings reveal a robust generic overgeneralization effect, predicted by the hypothesis that generics express primitive, default generalizations.