Endometrial hyperplasia (EH) is classified into benign and precancerous according to two different histomorphological systems: the World Health Organisation (WHO) system (based on the subjective evaluation of cytological atypia) and the endometrial intraepithelial neoplasia (EIN) system (based on a combination of several parameters that are assessable subjectively, or objectively through computerised analysis). The American College of Obstetricians and Gynecologists recommends use of the EIN system. Nonetheless, a higher prognostic value for EIN criteria was demonstrated only with the objective assessment, which is not routinely applicable. The aim of this study was to evaluate which of the subjective classifications of EH (WHO or EIN) has better prognostic value, by assessing the risk of coexistent cancer. Electronic databases were searched for relevant articles from the inception of the databases to July 2018. All studies assessing the presence of cancer on hysterectomy specimens after a preoperative histological diagnosis of EH were included. Odds ratios (ORs), sensitivity and specificity were calculated with 95% confidence intervals (CIs). Sixteen cohort studies and three case–control studies, assessing 2582 EHs, were included. The WHO criteria showed an OR of 11.15 (95% CI 7.65–16.24), a sensitivity of 0.86 (95% CI 0.82–0.90) and a specificity of 0.67 (95% CI 0.64–0.70) for coexistent cancer. The subjective EIN system showed a similar OR (11.85, 95% CI 4.91–28.62; P = 0.90), higher sensitivity (0.98, 95% CI 0.94–0.99), and lower specificity (0.29, 95% CI 0.24–0.34). The WHO system and the subjective EIN system have similar prognostic values. However, the EIN criteria appear to be more sensitive and thus more suitable for selecting women who need to be treated, whereas the WHO criteria, based on cytological atypia, seem to be more specific for lesions at higher risk of cancer. Therefore, integration of the EIN system with cytological atypia should be considered.