A critical component of the reinforcement model, incomplete postzygoti
c reproductive isolation, has been overlooked in the many selection ex
periments designed to test this model. Many field observations are con
sistent with the reinforcement model, but theoretical work predicts th
at reinforcement should only work when nearly complete postzygotic rep
roductive isolation has developed. The experiments described here used
four levels of postzygotic reproductive isolation: complete, strong,
moderate, and none. A significant decrease in heterotypic matings occu
rred over 25 generations in the complete postzygotic reproductive isol
ation treatment. No evidence for reinforcement was found in any of the
other treatments using incomplete postzygotic reproductive isolation.