The objectives of this study were to assess the association of a 4-point scale of vaginal discharge score (VDS) with time to pregnancy to define criteria for a practical case of purulent vaginal discharge (PVD) in dairy cows, to test the risk factors for PVD, and, finally, the effect of a dose of PGF2α on cure and reproductive performance. In experiment 1, grazing Holstein cows (n = 2,414) had their vaginal discharge scored at ∼32 d in milk (DIM) on a 4-point scale, the effect of VDS on the hazard of pregnancy by 300 DIM was then assessed to derive a case definition of PVD. Risk factors for PVD and self-cure were also assessed. In experiment 2, grazing Holstein cows (n = 6,326) from 5 herds were checked for PVD at ∼30 DIM. Cows with PVD were assigned to receive one dose of 500 µg of PGF2α analog (Cloprostenol; Ciclase, Syntex SA, Buenos Aires, Argentina) per cow (odd ear tag number) or to remain untreated (even tag number). Cure was declared if cows presented clear normal vaginal discharge (VDS-0) at visit 2 (∼62 DIM). Data were analyzed with Cox's regression and mixed logistic models. In experiment 1, cows with VDS ≥1 had lower hazard of pregnancy and longer calving to pregnancy interval than cows with VDS-0. This finding was not affected by the time at which the diagnosis was performed. Therefore, a cow ≥21 DIM and having VDS ≥1 was used to define a case of PVD. The odds of PVD were greater in primiparous cows compared with multiparous, in cows with abnormal calving compared with those with normal calving, and in those losing BCS peripartum. In experiment 2, PGF2α treatment tended to slightly increase the hazard of pregnancy (adjusted hazard ratio = 1.13). Conversely, PGF2α had no effect on the odds of cure of PVD [adjusted odds ratio (AOR) = 1.19], pregnancy at first service (AOR = 1.03), or pregnancy by 100 DIM (AOR = 0.89) or 200 DIM (AOR = 1.27). In conclusion, cows with VDS ≥1 can be considered to have PVD because of their lower hazard of pregnancy and longer calving to pregnancy interval (up to 48 d). Important risk factors are parity, calving, and body condition score loss peripartum. Optimal time of diagnosis is ≥28 to 35 DIM because cows experience a high self-cure rate. Self-cure is also affected by parity, prepartum BCS, and VDS. Finally, as treatment with one dose of PGF2α had a small effect on the hazard of pregnancy and no effect on clinical cure, its therapeutic use in grazing dairy cows with PVD is not recommended.