Ability to select service sires that minimize partial or complete losses of pregnancy could have major economic impacts in sheep production systems. This study tested the null hypothesis that survival of potential progeny did not vary with breed type of service sire or among individual rams. Data included 980 ewes on 10 farms; each ewe was pregnant to 1 of 67 rams of 12 breeds. Number of conceptuses was estimated once during pregnancy by ultrasonography, either transrectal (embryos) or transabdominal (fetuses), and was compared with number of lambs born to estimate losses. Data were examined first for number of lambs born and second for documented losses. Individual service sires affected number born (P < 0.001), which varied from 0.70 to 2.45 lambs per pregnant ewe. The main effects of breed type on lambs born were not significant, but breed types of both service sires (P < 0.0002) and ewes (P < 0.001) interacted with diagnosed number of conceptuses. Lambs born varied with ewe age (P < 0.0001) and among farms (P < 0.0001), and statistically, farms interacted with number of diagnosed conceptuses (P < 0.0001); season had no effect. In documented losses, there were both main effects of individual service sire and a service sire × number of diagnosed embryos interaction (P < 0.005). Thus, ewes bred to some rams were more apt to lose single pregnancies, whereas ewes bred to other rams were more apt to lose 1 or more embryos or fetuses from multiple pregnancies. Breed type of service sire affected (P < 0.05) prenatal death. Complete losses of single conceptuses tended to be greater in ewes bred to black-faced or hair-type rams (service sire breed type × number of diagnosed conceptuses; P < 0.09). Breed type of ewes also varied in incidence of complete losses (P < 0.05); hair-type ewes (46%) lost more (P < 0.02) documented conceptuses from examination to birth than black-faced (27%), white-faced (20%), or dairy-type (25%) ewes. Greater losses of singles than of multiples occurred in black-faced (37% vs. 18%) and hair-type (64% vs. 27%) ewes than in other breeds (ewe breed type × number of conceptuses; P < 0.03) per ewe. Surprisingly, purebred conceptuses were lost less often (24%) than crossbreds (36.4%; P < 0.002). Selection of rams based on records of prenatal losses in ewes they serviced may be a method to decrease embryonic and fetal wastage. However, further study to determine repeatability of differences among service sires from year to year will be required.