Animals and plants defend themselves against a variable community of biological enemies. We argue that the effectiveness of allocation to defence (the success of defence per unit allocation) may be expected to decrease as the diversity of attack types increases, and asked how the optimal allocation to defence covaries with the effectiveness of defence. Variation in effectiveness links optimal defence to coevolutionary processes; the prime characteristic of coevolutionary interactions is that they promote and maintain genetic variation in both hosts and their enemies, leading to variation in the effectiveness of defence. We present a simple model suggesting that as effectiveness decreases, the fitness benefit of defence disappears. In other words, when effectiveness is low, the optimal strategy is to tolerate damage. As effectiveness increases, the optimal allocation flips rapidly from no-defence (tolerance) to high allocation to defence, and then decreases at a decelerating pace as effectiveness increases. We conclude that diversifying coevolution, as it covaries with the effectiveness of defence, constrains the evolution of optimal defence strategies and may be a very important component in determining the optimal allocation to defence and variation in the success of defence as it is seen in the wild.