Perinatal health was investigated by linkage with the Medical Birth Registry of Norway for 192,417 births that took place between 1967 and 1991 among parents identified as farm holders in Norwegian agricultural censuses in 1969-1989. In a comparison with 61,351 births to nonfarmers in agricultural municipalities, farmers' births had an advantageous distribution of gestational ages and birth weights. Perinatal mortality was similar in the two groups, but the proportion of late-term abortions (gestational weeks 16-27) was higher among farmers' birth (odds ratio (OR) = 1.9, 95% confidence interval (CI) 1.6-2.3). Exposure indicators were classified on the basis of information given in the agricultural censuses and climate data for the grain harvest seasons of 1966-1991. The main hypotheses were that perinatal death is associated with parental exposure to pesticides. Toxoplasma contracted from infected sheep or pigs, or mycotoxins found in grain farming. There was no convincing evidence that perinatal death is associated with use of pesticides, sheep farming, or pig farming. The increase in late-term abortion among the farmers could to some extent be attributed to an excess of midpregnancy (weeks 21-24) deliveries among grain farmers; grain farmers had 132 deliveries at this time in pregnancy (2.8 per 1,000 pregnancies), while the nongrain farmers had 236 deliveries in midpregnancy (1.8 per 1,000). The authors found odds ratios (95% CI) that indicated that grain farming risk was higher after the harvest (1.8, 1.1-2.8), in seasons with a poor quality harvest (2.4, 1.5-3.8), and in pregnancies with multiple births (3.8, 1.7-8.2). These results support the hypothesis that occupational exposure to mycotoxins in grain induces labor at an early stage of pregnancy.