According to the Counterfactual Comparative Account of harm and benefit, an event is overall harmful (/beneficial) for a subject to the extent that this subject would have been better (/worse) off if it had not occurred. In this paper we present a challenge for the Counterfactual Comparative Account (CCA). We argue that if physical processes are chancy in the manner suggested by our best physical theories, then CCA faces a dilemma: If it is developed in line with the standard approach to counterfactuals, then it delivers that the value of any event for a subject is indeterminate to the extreme, ranging from terribly harmful to highly beneficial. This problem can only be avoided by developing CCA in line with theories of counterfactuals that allow us to ignore a-typical scenarios. Doing this generates a different problem: when the actual world is itself a-typical we will sometimes get the result that the counterfactual nonoccurrence of an actual benefit is itself a benefit. An account of overall harm bearing either of these two implications is deficient. Given the general aspiration to account for deprivational harms and the dominance of the Counterfactual Comparative Account in this respect, theorists of harm and benefit face a deadlock.