Recent studies of the relationship between progression from mild cognitive impairment (MCI) to Alzheimer disease (AD) and APOE ε4-allele revealed inconsistent results. To estimate the risk posed by APOE ε4-allele for developing AD in MCI subjects using meta-analysis and identify possible sources of heterogeneity between studies, we reviewed longitudinal epidemiological studies of the presence of APOE ε4-allele on risk for progression to dementia in MCI subjects, and conducted meta-analyses of the results from these studies. Our study was derived from 315 positive-outcome events and 461 negative-outcome events from 8 prospective studies. The pooled RR was statistically significant (pooled RR = 2.09; 95 % CI, 1.52-2.88). The Q statistics indicated high heterogeneity across studies (Q = 14.21, p = 0.003). RR were significantly related to the ethnicity of the sample (z = 3.58, p = 0.024). No significant heterogeneity was observed after stratification in four European-population studies (χ² = 0.67, p = 0.880), but it remained in four American-population studies (χ² = 18.52, p = 0.003). Heterogeneity markedly reduced after excluding one specific study (Q = 4.9, p = 0.009; I² = 39 %). APOE ε4-allele conferred increased risk for progression to dementia in MCI subjects. Ethnicity is a critical factor to yield heterogeneity. Further studies using larger sample sizes are required.