Assessing the efficacy of UN involvement in intemational crises may be made more difficult by the presence of selection effects. Selection effects introduce a sample bias that thwarts objective empirical assessment. The UN may only select easy or inexpensive crises or it may select the most severe and intractable crises. To compensate for selection effects, I employ a Heckman model to test for the detem1inants of successful UN crisis abatement. In addition to overcoming selection effects, the Heckman method also simulates a realistic sequential decision process. T he first step of the process models the initial UN decision to enter into a crisis. The second step models the detenninants of success after censoring crises in which the UN did not become involved. Failing to model the two steps as nonindependent processes can produce inconsistent, non-reliable and biased estimates in the second equation. I estimate jointly the UN decision to become involved in a crisis and the effectiveness of that involvement. Observatio ns are international crises from 1945 to 1994 identified by the lntemational Crisis Behavior ( ICB) data set. UN involvement is largely a positive function of the duration of the crisis, crisis gravity, and the number of actors involved. UN success is a positive fimction of mediation, calls for action, emergency forces, and observers. C risis violence undennines success. I discuss findings that suggest the factors that get the UN involved actually m ake it harder to achieve success.