Postpartum endometritis is the main cause of infertility in dairy cows, but there is a lack of critical diagnostic criteria. We aimed to 1) determine the optimal diagnostic method and criteria of endometritis for various postpartum days to evaluate infertility, and 2) assess the diagnostic accuracy of a combination of diagnostic methods. Holstein dairy cows (n = 441) from nine commercial dairy herds were examined at 42 ± 7 days postpartum by using 5 methods: 1) transrectal palpation measurement of the cervical diameter, 2) ultrasonographic measurement of the fluid in uterus (FIU) score, 3) vaginoscopic detection of external uterine orifice hyperemia, 4) vaginal discharge score (VDS), and 5) endometrial cytological percentage of polymorphonuclear leukocytes (PMN%). The clinical findings that were significant in the Pearson chisquare test and had the highest sum of sensitivity and specificity for infertility at 100, 125, 150, 175, and 200 days after parturition were determined the optimal criteria of endometritis. Logistic regression and the Cox proportional hazards model were used to compare the accuracies of the different diagnostic methods for infertility at various postpartum days. The combinations of methods which were significant in Pearson chisquare test and had the highest sum of sensitivity and specificity for infertility were proposed as the optimal combination for determination of endometritis status for various postpartum day. The optimal diagnostic criteria were PMN% ≥ 6.0, FIU ≥ 2 (continuous hyperechogenic or large amount of storage material), or VDS ≥ 2 (mucopurulent or worse vaginal discharge) for postpartum days 100 and 125; PMN% ≥ 8.0, FIU = 3 (large amount of storage material), VDS ≥ 2, or external uterine orifice hyperemia for day 150; PMN% ≥ 6.0, FIU = 3, VDS ≥ 2, or external uterine orifice hyperemia for day 175; and PMN% ≥ 5.0, FIU = 3, VDS ≥ 2, or external uterine orifice hyperemia for day 200. Only the results of endometrial cytology were related to infertility regardless of the postpartum days to evaluate infertility or statistical models. Compared with the sensitivity and specificity of a single diagnostic method, the sensitivity of a combination of methods improved but specificity decreased. We concluded that different diagnostic methods and criteria were required for postpartum days to evaluate infertility and diagnostic accuracy was improved by a combination of diagnostic methods rather than by a single method.