Granulocyte-macrophage colony-stimulating factor (GM-CSF) has been identified as a potentially important mediator of intercellular communication in the female reproductive tract, with principal target cells being the large populations of myeloid leukocytes in the cycling and pregnant uterus, the preimplantation embryo, and trophoblast cells of the developing placenta. To determine the physiological significance of this cytokine in reproduction, the fertility of genetically GM-CSF-deficient (GM-/-) mice was examined. Implantation rates were normal in GM-/- mice, and viable pups were produced. However, the mean litter sizes of GM-/- x GM-/- breeding pairs were 25% smaller at weaning than those of GM+/- x GM+/pairs, due to fetal death late in gestation and early in postnatal life, with a disproportionate loss of male pups. On Day 17 of pregnancy, the mean number of resorbing and malformed fetuses was twice as high in pregnant GM-/- females (21%, vs. 11% in GM+/- females); the mean fetal weight and the mean fetal:placental ratio in surviving conceptuses were diminished by 7% and 6%, respectively; and the number of very small fetuses (< 500 mg) was 9-times as high (23% vs. 2.5%). Mortality during the first 3 wk of life was 4.5-times as high in pups born to GM-/- mothers (9%, vs. 2% in GM+/- females), and diminished size persisted in GM-/- pups, particularly males, into adulthood. The detrimental effect of maternal GM-CSF deficiency was less apparent when GM-/- females were mated with GM+/+ males; litter sizes at birth and at weaning were not significantly smaller than in GM+/- matings, and fetal weights and fetal:placental ratios were also comparable. When polymerase chain reaction was used to genotype embryonic tissue in heterozygote matings, GM-/- fetuses from GM-/- females were found to be smaller than their GM+/- littermates and smaller than GM-/- fetuses gestated in GM+/- females. The size and distribution of uterine granulocyte and macrophage populations were normal during the estrous cycle, during early pregnancy, and in midgestation. Analysis of placental structure revealed that the ratio of labyrinthine to spongiotrophoblast areas was reduced by approximately 28% in GM-/- placentae, and the proportion of vacuolated trophoblast 'glycogen cells' in the spongiotrophoblast layer was diminished. Compromised placental function as a result of subtle developmental aberrations may therefore partially account for embryonic growth retardation in GM-CSF-deficient mice. Collectively, these studies show that fetal growth and viability are jeopardized in the absence of maternal GM-CSF. The detrimental effects are most clearly evident when the conceptus is also GM-CSF deficient, suggesting that GM-CSF of either maternal or fetal origin is required for optimal growth and survival of the fetus in mice.