In recent years, the role of PET imaging in the prediction of mild cognitive impairment (MCI) to Alzheimer’s disease (AD) conversion has been the subject of many longitudinal studies. The purpose of this study was to perform a meta-analysis to estimate the diagnostic accuracy of 18F-fluoro-2-deoxyglucose-positron emission tomography (FDG-PET) and 11C-Pittsburgh Compound B-positron emission tomography (PIB-PET) for prediction of short-term conversion to AD in patients with MCI. The MEDLINE and EMBASE databases were systematically searched for relevant studies. Methodological quality of the included studies was assessed. Sensitivities and specificities of PET in individual studies were calculated and meta-analysis was undertaken with a random-effects model. A summary receiver operating characteristic (SROC) curve was constructed with the Moses–Shapiro–Littenberg method. Heterogeneity was tested, and the presence of publication bias was assessed. Potential sources for heterogeneity were explored by assessing whether or not certain covariates significantly influenced the relative diagnostic odds ratio (DOR). Pooled estimates of sensitivity, specificity, positive likelihood ratio (LR+), negative likelihood ratio (LR–), DOR and the SROC curve of each PET imaging were determined. A total of 13 research studies (seven FDG-PET and six PIB-PET) met inclusion criteria and had sufficient data for statistical analysis. FDG-PET pooled estimates had 78.7% sensitivity (95% CI, 68.7–86.6%),74.0% specificity (95% CI, 67.0–80.3%), 18.1 LR+(95% CI, 7.3–45.0) and 0.32 LR–(95% CI, 0.16–0.61); and PIB-PET pooled estimates had 93.5% sensitivity (95%CI, 71.3–99.9%), 56.2% specificity (95% CI, 47.2–64.8%), 2.01 LR+ (95% CI, 1.57–2.58) and 0.17 LR–(95% CI, 0.08–0.36). Overall DOR was 17.3 (95% CI, 5.08–59.2) for FDG-PET and 12.8 (95% CI, 5.35–30.54) for PIB-PET. Area under the SROC curve was 0.88 ± 0.05 for FDG-PET and 0.85 ± 0.04 for PIB-PET. The data from FDG-PET research studies had high heterogeneity and funnel plot suggested a publication bias. The diagnostic accuracy determined for both FDG-PET and PIB-PET in this meta-analysis suggests that they are potentially valuable techniques for prediction of progression in patients with MCI. Both have their advantages and their combined use is a promising option for prediction purposes depending on availability and experience.