Breeding trials were designed to determine the influence of the mare on serum concentrations of equine chorionic gonadotropin (eCG) from Day 39 to Day 104 of gestation. Sires were ranked according to mean eCG concentrations found in the groups of randomly selected mares to which they were mated in 1983. Mares were ranked according to their mean eCG concentrations on Days 55, 71 and 85 of gestation (Day 0 = mating), in 1983 and 1985. In the 1986 breeding season, mares that had pregnancies characterized by high eCG levels were mated to sires previously associated with low eCG concentration pregnancies and low producing mares were mated to sires associated with high eCG concentration pregnancies. The highest eCG concentrations (12.8 ug/ml serum) were detected on Day 55 of gestation in mares mated to the low-ranked sire (P<0.01), indicating an influence of the mare on serum eCG concentration. A comparison of eCG production by individual mares in 1985, 1986 and 1987 showed that mares retained the same rank (P<0.001), regardless of the rank of the sire to which they were mated. It was concluded that the influence of the mare predominates in determining eCG concentration. In 1987, the highest ranked sire was mated to the highest eCG producing mares (ranked 1–20) and the second highest ranked sire was mated to mares ranked 21–40. Two distinct subgroups resulted, suggesting that the selection of sires and mares could be used to maximize the production of eCG.