The attack rate by the parasitoid Apanteles glomeratus (L.) on laboratory-reared larvae of Pieris rapae (L.) exposed for 3–4 day periods in a kale field in Massachusetts was greater when five larvae rather than two were placed on plants. The attack rate was greatest if host larvae were in the first instar when placed in the field, and little or no attack occurred when second- or third-instar larvae were exposed. The co-occurrence of unoccupied P. rapae larval feeding sites on plants where larvae were exposed reduced attack rates. Parasitism rates in trap-host larvae and in field-collected larvae (counting only cases in which the parasitoid was still in the egg stage) did not differ as estimators of recruitment to the population of parasitized hosts either in terms of total recruitment achieved per host generation or in the temporal pattern of the recruitment. In both simple and multiple linear regressions, the density of young P. rapae larvae (either parasitized or not) was the strongest correlate of parasitoid recruitment for the whole season, whereas average air temperature was the strongest correlate in the last month of observations (September).