Loss of P 27 expression correlates with clinical progression in a variety of human cancers. However, the correlation between P 27 expression and gastric cancer remains controversial. In this meta-analysis, we performed an electronic search based on six databases to select a sufficient number of studies. Pooled hazard ratio (HR) was used as estimates to investigate the association between P 27 expression and prognosis of patients with gastric cancer. We identified 19 studies with 2387 gastric cancer patients, ranging between 50 and 316 samples per study. Q and I 2 tests demonstrated that the homogeneity among 19 studies (I 2 = 47%, P = 0.0004), thus we applied a fixedeffects model to calculate the pooled HR of P 27 expression and overall survival (OS) of gastric cancer patients was 0.68, and 95% confidence interval (CI) was 0.60-0.78. Next, we conducted a subgroup meta-analysis and found that patients with low P 27 expression in Asians (HR = 0.69, 95% CI: 0.58-0.82) and non-Asians (HR = 0.57, 95% CI: 0.41-0.79) had poor prognosis. In addition, we found the publication bias results of OS in the final included 19 studies showed that this funnel plot presented incomplete symmetry, and then removed three literatures with larger HRs bias, and found that the remaining 16 literatures were homogeneity (I 2 = 0%, P = 0.47), the pooled HR was 0.52 with 95% CI of 0.43-0.62, and the publication bias disappeared. These results suggested a strong association between P 27 underexpression and poorer prognosis of gastric cancer in patients. P 27 may be a tumor suppressor for predicting survival outcome of gastric cancer patients. Anti-Cancer Drugs 33: e692-e699