The claim that the spread of nuclear weapons leads to interstate conflict and nuclear war has become very influential. However, proliferation pessimists have failed to specify how and when nuclear proliferation precipitates conflict. I make four arguments for an optimistic pessimism. (1) The few preventive strikes against nuclear facilities that have occurred would have occurred absent of the target's nuclear program, and these rare strikes did not lead to conflict escalation. (2) The problem of nonsurvivable arsenals is, properly understood, a problem of preventive-war motivations where subjective uncertainty reduces the dangers of arsenal survivability. (3) Claims that bias within nuclear organizations may lead to accidental nuclear detonations suffer from omitted variable bias: leaders' decisions to revise the status quo after developing nuclear weapons tend to give rise to the most dangerous nuclear accidents. Accidents that have not occurred during a nuclear crisis pose substantially less risk of nuclear escalation. (4) Leaders of nuclear states have tended to engage in conventional aggression, but experience with nuclear weapons moderates their conflict propensity. Ultimately, I argue that while nuclear weapons have led to conflict through one causal mechanism and for a limited time, the dangers are substantially weaker than usually assumed.