Deficits in decision-making using the Iowa Gambling Task (IGT) have been found in past suicide attempters, but primarily euthymic and/or medicated patients. This study compared IGT performance among medication-free, currently depressed patients (unipolar and bipolar) with (n = 26) and without (n = 46) a past history of suicide attempt, and healthy volunteers (n = 42). Attempter status, in a sample whose attempts were predominantly non-violent, was not associated with impaired IGT performance even when accounting for sex, mood disorder type, and comorbid Borderline Personality Disorder. A non-significant trend towards poorer performance was found in a small subgroup of past attempters who had used a violent method, consistent with prior studies. Suicide intent and ideation were unrelated to IGT scores. There were no consistent associations between IGT performance and ratings of impulsiveness (Barratt Impulsiveness Scale (BIS)), hostility (Buss-Durkee Hostility Inventory (BDHI)) or aggression (Brown-Goodwin Aggression Inventory (BGAI)). Results suggest that decision-making impairment is related to specific subtypes of suicidal behavior, but may not be universally sensitive to suicide risk in all types of attempters, especially those using non-violent means. Psychometric and conceptual issues surrounding the IGT also appear to affect its utility as a general marker of suicidal behavior risk.