Curcumin is a polyphenolic natural compound with diverse and attractive biological activities, which may prevent or ameliorate pathological processes underlying age‐related cognitive decline, dementia, or mood disorders. However, clinical trials and animal studies have yielded conflicting conclusions regarding its effectiveness for cognition in different individuals. The aim of this review is to meta‐analytically assess the effectiveness of curcumin for cognitive function in different types of people. A preliminary search on PubMed, Embase, Web of Science, ClinicalTrials.gov, Cochrane Library, Chinese National Knowledge Infrastructure, and Wanfang Data and China Biology Medicine disc was performed to identify randomized controlled trials investigating the effect of curcumin on cognition. Six clinical trials with a total of 289 subjects met inclusion criteria for this review. We used a random‐effects model to calculate the pooled standardized difference of means (SMD). For older adults who received curcumin, scores on measures of cognitive function (SMD = 0.33, 95% confidence interval [CI] [0.05, 0.62]; p = 0.02), occurrence of adverse events (odds ratio [OR] = 5.59, 95% CI [0.96, 36.80]; p = 0.05), and measures of depression (SMD = −0.29, 95% CI [0.64, 0.05]; p = 0.09) indicated significant memory improvement. In patients with Alzheimer's disease (AD), scores in measures of cognition status (SMD = −0.90, 95% CI [1.48, −0.32]; p = 0.002) indicated that there was a trend for treated subjects to do worse than placebo‐treated subjects on the Mini‐Mental State Examination. The occurrence of adverse events (OR = 0.87, 95% CI [0.10, 7.51]; p = 0.90) was similar to those who received placebo. Due to insufficient data, it was impossible to provide a narrative account of only the outcomes for schizophrenia. Curcumin appears to be more effective in improving cognitive function in the elderly than in improving symptoms of AD and schizophrenia. Curcumin is also safe and tolerated among these individuals. Because of the small number of studies available, a funnel plot or sensitivity analysis was not possible. Further high‐quality trials with larger sample sizes or bioavailability‐improved curcumin formulations may be considered for reliable assessment.