Introduction: Evidence suggests that disturbed sleep may constitute an important and modifiable vulnerability factor for acute suicidal risk and completed suicide. Sleep complaints are closely coupled with mood problems, and depression is considered the single best predictor of suicide. A paucity of research has investigated how specific indices of sleep may predict suicide risk, beyond depressive symptoms, using an objective measurement of sleep and a prospective study design. Methods: Data were collected among 49 men and women across three study time points: baseline (T1), 1 week (T2), and 3 weeks (T3). Using wrist actigraphy, objective indices of sleep were hypothesized to predict increases in suicidal symptoms over time, even after controlling for depression. The Beck Depression Inventory (BDI) and Beck Scale for Suicide (BSS) were used as primary symptom measures. We furthermore hypothesized that mood lability, versus depressed mood per se, would account for the relationship between sleep and suicidality. Results: Hypotheses were partially supported. Regression 1 indicated that, as a set, actigraphy sleep variables, significantly predicted T2 residual changes scores, controlling for BDI (P < .001). Specifically, Sleep variability (P < .05) uniquely predicted T2 BSS residual change scores, whereas sleep efficiency, WASO (Wake After Sleep Onset), sleep latency, and sleep duration did not (P > .05). Regression 2 results also indicated that actigraphy sleep variables together predicted T3 BSS change scores, controlling for BDI (P < .001). Again, higher sleep variability uniquely predicted T3 BSS change scores, though as a nonsignificant trend (P = .06). Sleep efficiency, WASO, sleep latency, and sleep duration (P > .05) did not significantly contribute to the variance beyond that contributed by sleep variability. A Sobel test revealed that depression scores significantly mediated the relationship between sleep variability and BSS T2 change scores (Z = 2.03, P = .04); this finding was partially replicated as a nonsignificant trend for BSS T3 change scores (Z = 1.28, P = .19). In contrast, a mood lability mediation model was not supported according to standard tests of mediation (P > .05 for BSS T2 and T3 residual change scores, respectively). Conclusions: Actigraphically-assessed sleep variability was uniquely associated with greater levels of suicidal ideation, an effect that was independent of depression. Low mood, versus increased mood lability, mediated the relationship between sleep and increases in suicide risk. Clinical implications are discussed.