Based on previous research, we hypothesized that Cosynch at 72 h [GnRH-7 d-PGF(2alpha)-72 h-GnRH + artificial insemination (AI)] would result in a greater number of pregnancies per AI (P/AI) than Cosynch at 48 h. Further, we hypothesized that P/AI would be improved to a greater extent when GnRH was administered at 56 h after PGF(2alpha) before AI at 72 h due to a more optimal interval between the LH surge and AI. Nine hundred twenty-seven lactating dairy cows (n = 1,507 AI) were blocked by pen, and pens rotated through treatments. All cows received GnRH followed 7 d later by PGF(2alpha) and then received one of the following: 1) GnRH + timed AI 48 h after PGF(2alpha) (Cosynch-48); 2) GnRH 56 h after PGF(2alpha) + timed AI 72 h after PGF(2alpha) (Ovsynch-56); or 3) GnRH + timed AI 72 h after PGF(2alpha) (Cosynch-72). Pregnancy diagnoses were performed by ultrasound at 31 to 33 d post-AI and again at 52 to 54 d post-AI. Overall P/AI were similar for the Cosynch-48 (29.2%) and Cosynch-72 (25.4%) groups. The Ovsynch-56 group had a greater P/AI (38.6%) than Cosynch-48 or Cosynch-72. Presynchronized first-service animals had greater P/AI than cows at later services in Cosynch-48 (36.2 vs. 23.0%) and Ovsynch-56 (44.8 vs. 32.7%) but not in Cosynch-72 (24.6 vs. 26.2%). Similarly, primiparous cows had greater P/AI than multiparous cows in Cosynch-48 (34.1 vs. 22.9%) and Ovsynch-56 (41.3 vs. 32.6%), but not Cosynch-72 (29.8 vs. 25.3%). In conclusion, we found no advantage to Cosynch at 72 h vs. 48 h. In contrast, we found a clear advantage to treating with GnRH at 56 h, 16 h before a 72-h AI, probably because of more-optimal timing of AI before ovulation.