Black mustard, Brassica nigra, is highly variable in both constitutive resistance and induction response following damage by herbivores. A focal population from Ithaca, New York, was used to test the following two predictions of optimal defense theory: (1) that allocation to resistance will reduce plant performance in the absence of herbivores; and (2) that induction response will be negatively correlated with constitutive resistance. The experiment consisted of a half-sib mating design with 47 paternal families and four dams per sire, fully crossed with a damage treatment consisting of 25% leaf removal by the cabbage white butterfly, Pieris rapae, when plants had four leaves. Leaf trichome density, sinigrin concentration, and glucobrassicin concentration were 38%, 19%, and 16% higher, respectively, for the seventh leaf of damaged plants. Paternal families did not vary significantly in their induction response. Narrow-sense heritabilities were ϭ 0.51, 0.76, and 0.50 for constitutive leaf trichome density, sinigrin 2 h s concentration, and glucobrassicin concentration, respectively. Positive genetic correlations were found between glucobrassicin concentration and days to first flower, suggesting a genetic cost of resistance. Induction responses were negatively correlated with constitutive allocation for leaf trichome density and sinigrin concentration. The results were therefore consistent with optimal defense theory, offering modest evidence for both predictions.