The problem of outcome responsibility, in criminal law, is the problem of providing a normative foundation for criminal-law practices that make outcomes relevant to criminal liability and sentencing – sometimes (as in the law of attempts) with relevant differences between intended and negligently-risked outcomes. My talk develops a conception of agency that can provide theoretical underpinning for these practices. As a first step, I defend the widely accepted view that producing an outcome (either as an intended or as a negligently-unlucky one) makes a difference: so long as the outcome can be imputed to the agent, it qualifies what the agent is responsible for. The object of assessment (“what D has done”) changes from “X activity” (minus the outcome) to “X activity with outcome”. Potentially much more contentious is the second step in my argument: the claim that how the outcome matters for our responsibility is affected by the question whether the outcome is the product of our intentions or whether it is the product of risks we have negligently created. This analysis explains many contentious features of the criminal law. However, it puts pressure on the justification for outcome-dependent crimes of negligence as such. For how can, on the assumption that we have a much less tight connection to negligently-risked outcomes than to intended ones, the criminal law’s practice of making the occurrence of the outcome the linchpin for most offences of criminal negligence, be defended?