Cardiac magnetic resonance (CMR) has been recognized as the gold standard for the evaluation of left ventricular (LV) function. Cardiac gated PET allows the simultaneous assessment of LV function with the evaluation of myocardial perfusion and metabolism. But the correlations between PET and CMR remain controversial. We conducted a systematic electronic search of PubMed, Embase and the Cochrane Library . Forest plot, spearman correlation analysis and Bland-Altman analysis were used to evaluate the correlations between PET and CMR. Pooled analysis of 13 studies showed that PET underestimated left ventricular end-diastolic volumes (LVEDV) [mean difference (MD), -15.30; 95% confidence interval (CI), -23.10 to -7.50; P < 0.001] and left ventricular end-systolic volumes (LVESV) (MD, -6.20; 95% CI, -12.58 to 0.17; P = 0.06) but not left ventricular ejection fraction (LVEF) (MD, -0.35; 95% CI, -1.75 to 1.06; P = 0.63). Overall, there were very good correlations between PET and CMR measurements for LVEDV ( r , 0.897), LVESV ( r , 0.924) and LVEF ( r , 0.898). Subgroup analysis indicated that LVEDV ≥180 ml and LVEF <40% reduced the accuracy of PET, especially the measurement of LVEF ( r , LVEDV ≥180 vs . r , LVEDV < 180 : 0.821 vs. 0.944; r , LVEF < 40% vs . r , LVEF ≥40% : 0.784 vs. 0.901). Correlations between PET and CMR measurements of LVEDV, LVESV and LVEF were excellent, but these two methods could not be used interchangeably for accurate measurements of LV volume and LVEF in patients with significantly increased LV volume and decreased LVEF.