none Optimal clutch size theory predicts that for similar species of parasitoids, smaller hosts should impose more severe density-dependent competition than larger hosts and that, consequently, selection should favour smaller clutches in smaller hosts. These predictions were tested by comparing the strength of density dependence experienced by a native population of Cotesia glomerata attacking a large host species, Pieris brassicae , in Britain, with that experienced by a population of C. glomerata , introduced over 100 years ago into the U.S.A., attacking a smaller host species, P. rapae . No density dependence of adult size, survival or development time was found for the native population attacking P. brassicae . In contrast, strong density dependence of adult size and survival and found for the introduced population attacking P. rapae . Contrary to expectation, the mean clutch size laid by both the native and introduced populations of C. glomerata in P. rapae (26 multiplied by 6 and 28 multiplied by 0 eggs, respectively) were significantly larger than that laid by the native population in P. brassicae (20 multiplied by 4 eggs). The mean clutch size laid by the introduced population in P. rapae was also larger than that predicted to maximize fitness per host.