Abstract.1We examine the conventional wisdom among biological control practitioners that destructive host feeding is a desirable attribute in parasitoids employed for classical biological control, using both the predictions of population dynamics theory and historical data on biological control introductions of Hymenoptera against Homoptera.2Population dynamics theory predicts that destructive host feeders, compared with other parasitoids, are (a) either just as likely or more likely to become established, and (b) unable to depress host equilibria as strongly.3Analyses of the B IOC AT database suggest that among parasitoids of Homoptera destructive host feeders are superior to other parasitoids with respect to both establishment rate and success rate.4We present likely explanations for the disparity between the predictions of population dynamics theory and the results of database analysis. A partial explanation for the mismatch between theory and observation with respect to the degree of pest suppression may be that females of destructive host feeding parasitoids rely less upon hosts as a food source when alternative foods such as honeydew and nectar are plentiful.5We conclude that, despite the predictions of population dynamics theory, destructive host feeders are probably better biological control agents than other parasitoids, and certainly no worse, but that it would imprudent to use destructive host feeding as the sole, or even primary, selection criterion when seeking agents for classical biological control.