The effect of altitude on birth weight was measured with data in U.S. natality records from 1978 to 1981 after correlation with the mean altitude of the mother's resident county. For comparison of the low birth weight (LBW) rate at different altitudes, certain socioeconomic risk factors known to affect birth weight were controlled by the selection of an idealized subpopulation of singleton births. With 500 m gradations for altitude, a curvilinear dose-response relationship of birth weight reduction with increasing altitude was demonstrated (P less than 0.001). In comparison with neonates born at sea level, neonates born at higher altitudes (greater than 2000 m) had a twofold to threefold increase in LBW rate, mainly related to a higher incidence of intrauterine growth retardation. Comparison of the LBW rate on the basis of small geographic divisions in the mountain states showed a positive correlation between the LBW rate and the high altitude. The birth weight frequency distribution curves of the idealized subpopulation at each altitude approximate normal distributions and parallel one another, indicating that altitude has a general effect on all births. Such nearly normal birth weight distributions allowed the determination of altitude-specific LBW cutoff limits that can be used to detect areas of greater risk for LBW but independent of the altitude effect.