The purpose of this study was to examine patterns of self-reported suicidality and distress during research assessments in a sample of 63 women meeting criteria for borderline personality disorder and current and chronic suicidality. The risk management protocol we used during the two-year study period (University of Washington Risk Assessment Protocol; UWRAP) is described. Results indicated that changes in suicidality following assessments were small and relatively infrequent, and were just as likely to reflect decreases in suicidality as increases (17.5% versus 16.4% of sessions, respectively). Further, longitudinal analyses indicated that changes in suicidality became increasingly rare over the course of the 2-year study. Ratings of distress were more changeable than suicidality, underscoring the need for separate measurement of these constructs when assessing risk. With the aid of the UWRAP, our assessors judged 15 participants as high-risk status in 28 assessment sessions (3.7% of all sessions). In comparison to the rest of the sample, these individuals were of significantly greater clinical severity as measured by the HRSD 17-item, GAF scores, number and severity of previous suicide attempts, and number of previous psychiatric hospitalizations. Low-intensity risk intervention strategies (e.g., validating participant's feelings) were typically sufficient to reduce risk in these participants. Overall, our findings indicate that research with highly suicidal individuals can be done safely with the use of well-trained assessors and an appropriate crisis management protocol.