Data from 257 normally cyclic mares used in trials over a 10-year period were examined to establish the relationship between the number of inseminations per cycle, duration of oestrus and pregnancy rate. Mares that became pregnant were inseminated more (P less than 0.05) frequently per cycle than mares that did not become pregnant (3.4 versus 2.8). First-cycle pregnancy rates of 22.2, 34.0, 38.6, 52.5, 58.3 and 52.2% were attained when mares were inseminated 1, 2, 3, 4, 5 or 6 or more times per cycle, respectively. The duration of oestrus in mares that became pregnancy was longer (P less than 0.01) than in those that did not conceive (7.9 versus 6.4 days). Studies of 64 mares in the transitional season showed that first-cycle pregnancy rates for mares in which oestrus lasted less than 10, greater than or equal to 10, greater than or equal to 20, greater than or equal to 30 and greater than 40 days were 56.2, 76.2, 76.5, 77.3 and 88.9%, respectively. Overall pregnancy rates (after 3 cycles) were lower (P less than 0.05) for mares that had an initial oestrus of less than 10 days. Mares were inseminated every other day of oestrus with 100 X 10(6) progressively motile spermatozoa. First-cycle pregnancy rates were 64.3, 75.0 and 75.0% for mares inseminated 1-4, 5-10 and greater than or equal to 12 times per cycle, respectively. Fewer (P less than 0.05) mares became pregnant after 3 cycles when inseminated 1-4 times per cycle than did mares inseminated greater than or equal to 12 times per cycle (51.3 versus 75.0%). Numerous inseminations per cycle of mares with extended oestrus were not detrimental to fertility.