The present study examined the relationship between anxiety and premature termination, and the effect of depression and a proxy measure of impulsivity on this relationship. Extensive research has been conducted that has focused on the prediction of premature termination from psychotherapy, but many of the results have been inconclusive. One common result from the extant literature is that nearly half of therapy clients reportedly terminate from therapy prematurely, but there has been little consensus regarding which variables best predict dropout. Some evidence has suggested that more severe symptoms, such as anxiety, might predict persistence in therapy. Accordingly, it was hypothesized that clients with higher levels of physiological hyperarousal, a specific component of anxiety, would be less likely to terminate prematurely. This prediction was thought to be evident only in clients with low levels of social anxiety; higher physiological hyperarousal was predicted to increase odds of premature termination in clients with higher levels of social anxiety. In addition, it was hypothesized that clients with higher levels of depression would be more likely to terminate therapy prematurely. The possible moderating effect of a proxy measure of impulsivity on the hypothesized relationships was also examined. To test these hypotheses, hierarchical logistic regression analyses were applied to a sample of client data from a large university outpatient therapy clinic. Analyses were repeated for the total sample, high social anxiety group, and low social anxiety group. Although the results did not support the primary hypothesized relationship between physiological hyperarousal and premature termination, trends were present that suggested some limited support for the proposed predictive relationship between depression and premature termination, even after controlling for initial symptom severity. Limitations and recommendations for future premature termination research are discussed.