To assess the safety of galantamine in subjects with mild cognitive impairment (MCI), the ability of galantamine to benefit cognition and global functioning in subjects with MCI, and the ability of galantamine to delay conversion to dementia. In two studies, 2,048 subjects, 990 in Study 1 and 1,058 in Study 2, with a Clinical Dementia Rating (CDR) = 0.5, CDR memory score > or =0.5, without dementia were randomized to double-blind galantamine (16-24 mg/day) or placebo for 24 months. Primary efficacy endpoint at month 24 was number (%) of subjects who converted from MCI to dementia (CDR > or = 1.0). There were no differences between galantamine and placebo in 24-month conversion rates (Study 1: 22.9% galantamine vs 22.6% placebo, p = 0.146; Study 2: 25.4% galantamine vs 31.2% placebo, p = 0.619). Mean CDR-sum of boxes declined less with galantamine than placebo at 12 and 24 months in Study 1 (p = 0.024 12 months and p = 0.028 24 months), but not in Study 2 (p = 0.662 12 months and p = 0.056 24 months). Digit Symbol Substitution Test scores improved with galantamine in Study 1 at 12 months and in Study 2 at 24 months (Study 1: p = 0.009 month 12 and p = 0.079 Month 24; Study 2: p = 0.154 month 12 and p = 0.020 month 24). The most frequently reported adverse event was nausea (galantamine, 29%; placebo, 10%). Serious AEs occurred in 19% of each group. Mortality of the cohort after retrospectively determining the status of subjects (98.3%) at 24 months was 1.4% (galantamine) and 0.3% (placebo); RR (95% CI), 1.70 (1.00, 2.90). Galantamine failed to significantly influence conversion to dementia. Galantamine was generally well tolerated. Whereas recorded mortality was greater in the galantamine group than in the placebo group in the original per-protocol assessment, a post hoc analysis of the cohort was consistent with no increased risk.