Records from 16,351 litters born between 1991 and 1996 from Duroc, Landrace and Yorkshire sows and Landrace × Yorkshire crossbred sows were used to estimate the effects of year, parity and breed on eight sow productive traits at birth. The model used in this study included effects of year, parity, breed, sow nested within breed, and the interactions of year by parity and parity by breed. The results obtained were as follows: Year had highly significant effects (P$lt;0.01) on the numbers of stillbirths, pigs mummified, small pigs, and pigs suckled and litter weight. The breeds of sow significantly affected (P$lt;0.01) the total number of pigs born, the number of stillbirths, small pigs, pigs culled, and pigs suckled and litter weight. The total number of pigs born, the number of pigs suckled and litter weight were the highest in the 4th parity. The number of stillbirths, pigs mummified and small pigs tended to inaease with the inaease of parity. Parity effects were significant on most of the traits. The total number of pigs born and the number of pigs suckled were greater in the crossbred sows than in the purebreds (P$lt;0.01) and litter weight was heavier in the crossbred sows than in the purebreds (P$lt;0.05). In purebreds, Landrace and Yorkshire sows were superior to Duroc sows in the total number of pigs born and the number of pigs suckled (P$lt;0.01). Landrace Yorkshire crossbred sows were similar to Yorkshire Landrace crossbreed in the sow productive traits. Heterosis were 9.3% for the total number of pigs, 11.6% for the number of pigs suckled, 4.3% for litter weight, 25.0% for the number of stillbirth, 23.1% for the number of pigs mummified and 20.0% for the number of pigs culled.