High cortisol levels are found in severe mood disorders, particularly bipolar disorder. Hypercortisolaemia may cause or exacerbate both neurocognitive impairment and depressive symptoms. We hypothesized that antiglucocorticoid treatments, particularly corticosteroid receptor antagonists, would improve neurocognitive functioning and attenuate depressive symptoms in this disorder. To test this hypothesis, 20 bipolar patients were treated with 600 mg/day of the corticosteroid receptor antagonist mifepristone (RU-486) or placebo for 1 week in a double-blind crossover design. Over the total 6 weeks of the study, neurocognitive and neuroendocrine function were evaluated at baseline, days 21 and 42. Mood symptoms were evaluated weekly. Nineteen subjects completed the protocol; there were no drop-outs due to adverse events. Following treatment with mifepristone, selective improvement in neurocognitive functioning was observed. Spatial working memory performance was significantly improved compared to placebo (19.8% improvement over placebo). Measures of verbal fluency and spatial recognition memory were also improved after mifepristone. Beneficial effects on mood were found; Hamilton Depression Rating Scale scores were significantly reduced compared to baseline (mean reduction of 5.1 points) as were Montgomery-Asberg Depression Rating Scale scores (mean reduction of 6.05 points). No significant change occurred after placebo. These data require replication but provide preliminary evidence that glucocorticoid receptor antagonists may have useful cognitive-enhancing and possibly antidepressant properties in bipolar disorder.