Direct and indirect (mediated by biologic factors) effects of inbreeding on prereproductive mortality (death before age 20 years) were investigated in a case-control study conducted in the Lancaster County, Pennsylvania, Old Order Amish. A total of 211 cases of prereproductive death between 1969 and 1980 and 213 live controls were compared for differences in inbreeding coefficients, congenital malformations, birth weight, gestational age, birth complications, and other demographic factors, obtained by linking cases and controls to vital records and the Amish genealogic registry dating back to the 1700s. Inbreeding coefficients (F) for cases and controls were computed using the path method of tracing common ancestors in the multigenerational pedigrees. Close inbreeding (F greater than or equal to 1/64) was a significant risk factor for prereproductive mortality; odds ratio = 1.55. Using the log-linear model, the effect of close inbreeding on mortality was found to be mediated by three indirect casual mechanisms: Regardless of case-control status, inbreeding was significantly related to congenital malformations (recorded at birth), intrauterine growth retardation (birth weight less than 10th percentile for gestational age), and the occurrence of other deaths in the sibship. In turn, each of these factors was independently related to mortality regardless of inbreeding. No significant direct effect of inbreeding remained after adjustment for these factors. There were no effects of inbreeding on prematurity (less than 37 weeks), or birth complications. This study suggests that inbreeding increases the risk of prereproductive mortality by increasing the risk of intrauterine growth retardation and congenital malformations but not prematurity. The log-linear model provides a useful approach to the analysis of direct and indirect risk factors using biologic mechanisms.