The state of the psychotherapy termination literature to date might best be characterized as inconclusive. Despite decades of studies, almost no predictors of premature termination have emerged consistently. An examination of this literature reveals a number of recurrent methodological-analytical problems that likely have contributed substantially to this state. Survival analysis, which was designed for longitudinal data on the occurrence of events, not only circumvents these problems but also capitalizes on the rich features of termination data and opens brand new avenues of investigation. A substantive illustration of survival analysis is provided through tests of 5 variables and their interaction with time in the prediction of both premature and mutual termination.