The birth prevalence of specific types of congenital anomalies at low and high altitudes in South America were compared after adjustment for prenatal growth, ethnicity, and socioeconomic status. The material includes all 1,668,722 consecutive births occurring in 53 hospitals participating in the Latin-American Collaborative Study of Congenital Malformations (ECLAMC), having registered at least 100 malformed/control pairs between 1967 and 1995. The lowland subsample (below 2,000 m above the sea level) included 46,729 case-control pairs, ascertained in 1,539,432 births from 49 hospitals in 38 cities. The highlands (above 2,000 m) comprised 3,498 case-control pairs from 129,301 births, occurring in four hospitals from three cities. Unconditional logistic regression was used to estimate the relative risks for the exposure at high altitudes, adjusted by ethnicity (Amerindian yes/no), type of health service (public/private), and birth weight (below/equal and greater than 2, 500 g). The adjusted relative risks showed significantly (P < 0.01) higher values in the high than in the lowlands for four types of defects: cleft lip [relative risks (RR): 1.57; 95% confidence interval (CI): 1.27-1.94], microtia (RR: 3.21; 95% CI: 2.35-4.79), preauricular tag (RR: 2.09; 95% CI: 1.86-2.36), branchial arch anomaly complex (RR: 1.79; 95% CI: 1.23-2.61), constriction band complex (RR: 1.92; 95% CI: 1.11-3.31), and anal atresia (RR: 1.61; 95% CI: 1.01-2.57). Conversely, lower risks in the highlands were registered for two neural tube defects: anencephaly (RR: 0.33; 95% CI: 0.20-0.54), spina bifida (RR: 0.57; 95% CI: 0.37-0.78), as well as for hydrocephaly (RR: 0.41; 95% CI: 0.22-0.77) and pes equinovarus (RR: 0.70; 95% CI: 0.51-0.96). Even though some of these differences may be caused by undetected confounders, the coincidental finding of four types of craniofacial defects with higher, and two types of neural tube defect with lower frequencies in the highlands, suggest a real biological foundation.