The aim of this study was to determine whether an increase in circulating estrogen concentrations would increase percentage pregnant per artificial insemination (PP/AI) in a timed AI protocol in high-producing lactating dairy cows. We analyzed only cows having a synchronized ovulation to the last GnRH of the Ovsynch protocol (867/1,084). The control group (n = 420) received Ovsynch (GnRH--7 d--PGF(2alpha)--56 h--GnRH--16 h--timed AI). The treatment group (n = 447) had the same timed AI protocol with the addition of 1 mg of estradiol-17beta (E2) at 8 h before the second GnRH injection. Ovarian ultrasound and blood samples were taken just before E2 treatment of both groups. In a subset of cows (n = 563), pressure-activated estrus detection devices were used to assess expression of estrus at 48 to 72 h after PGF(2alpha) treatment. Ovulation was confirmed by ultrasound 7 d after timed AI. Treatment with E2 increased expression of estrus but overall PP/AI did not differ between E2 and control cows. There was an interaction between treatment and expression of estrus such that PP/AI was greater in E2-treated cows that showed estrus than in E2-treated or control cows that did not show estrus and tended to be greater than control cows that showed estrus. There was evidence for a treatment by ovulatory follicle size interaction on PP/AI. Supplementation with E2 improved PP/AI in cows ovulating medium (15 to 19 mm) but not smaller or larger follicles. The E2 treatment also tended to improve PP/AI in primiparous cows with low (< or =2.5) body condition score, and in cows at first postpartum service compared with Ovsynch alone. In conclusion, any improvements in PP/AI because of E2 treatment during a timed AI protocol appear to depend on expression of estrus, parity, body condition score, and size of ovulatory follicle.