Despite the general recognition that low birth weight can be caused by many factors, confusion and controversy remain about which factors have independent causal effects, as well as the quantitative importance of those effects. Previous research findings have often been conflicting because of a failure to distinguish intrauterine growth retardation from prematurity, differences in focus (means v rates), inadequate control for confounding variables, and insufficient statistical power. This review of the English and French language medical literature published between 1970 and 1984 is based on a critical assessment and meta-analysis of 895 studies concerning 43 potential determinants of intrauterine growth or gestational duration. Based on methodologic standards established a priori for each candidate determinant, the best studies were used to assess the existence and magnitude of an independent causal effect on birth weight, gestational age, prematurity, and intrauterine growth retardation. Factors with well-established direct causal impacts are identified, and their relative importance is indicated for "typical" developing and developed country settings. Modifiable factors with large effects are targeted for public health intervention in the two settings. Finally, factors of potential quantitative importance, but for which data are either unavailable or inconclusive, are highlighted as priorities for future research.