Background <p>Existing suicide risk screening and detection methods are unable to reliably identify who will engage in suicidal behavior (SB), and when. Modeling the “push-pull” among risk and protective processes could be useful for understanding when patients may be at increased risk for transitioning from lower to higher risk states. This study examines if changing temporal dynamics in wish to live (WTL) and wish to die (WTD) ratings, assessed during consecutively scheduled psychiatric outpatient visits, can differentiate patients who will attempt suicide and signal when suicidal behavior is about to occur.</p> Methods <p>78 adults with past-month suicidal ideation and/or suicidal behaviors who enrolled in a randomized clinical trial comparing two outpatient psychotherapies provided weekly WTL and WTD ratings assessed with the first two items of the Scale for Suicide Ideation and weekly occurrence of suicide attempts assessed with items from the Self-Injurious Thoughts and Behaviors Interview-Revised.</p> Results <p>Session-to-session changes in WTL and WTD were examined using multilevel regression modeling supplemented by eigenvalue decomposition analysis. Temporal patterns significantly differed over time between patients who attempted suicide versus patients who did not. Temporal patterns remained consistent over time among patients who did not attempt suicide. Among patients who attempted, temporal patterns changed from pre- to post-attempt, consistent with a state transition.</p> Conclusions <p>Suicidal behaviors represent a state transition that is foreshadowed by unique temporal patterns in self-reported WTL and WTD.</p>