To compare 2 strategies for systematically resynchronizing ovulation, lactating Holstein cows (n = 763) at various days in milk and prior artificial insemination services were assigned randomly at timed AI (TAI) to receive the first GnRH injection of Ovsynch 26 (D26) or 33 (D33) d after TAI to resynchronize ovulation (Resynch) in cows failing to conceive. Cows in the D26 treatment received GnRH 26 d after TAI and continued Resynch only when diagnosed not pregnant by using ultrasonography 33 d after TAI, whereas D33 cows initiated Resynch only when diagnosed not pregnant 33 d after TAI. Cows were classified based on the presence or absence of a corpus luteum (CL) at the not-pregnant diagnosis, and cows without a CL received an intravaginal progesterone-releasing insert during Resynch. When analyzed as a systematic strategy, pregnancy rate per AI (PR/AI) was greater for cows assigned to the D33 than the D26 Resynch treatment (39.4 vs. 28.6%). A treatment x parity interaction was detected for PR/AI after Resynch for nonpregnant cows having a CL in which primiparous cows had a greater PR/AI than multiparous cows when Resynch was initiated 33 d after the initial TAI, and primiparous and multiparous cows when Resynch was initiated 26 d after the initial TAI. Pregnancy loss for Resynch was 6.4% between 33 and 40 d, and 2.6% between 40 and 61 d after Resynch TAI. We concluded that delaying initiation of Resynch until 33 d after TAI increased PR/AI for primiparous cows.