A case-control study on work with video screen equipment during pregnancy has been made for three cohorts of women, identified with the aid of occupational codes in the census, linked to the Medical Birth Registry and a registry containing information on women hospitalized for spontaneous abortion. Five hundred and twenty-two cases were selected (women with spontaneous abortions or women who had infants that died, had severe malformations, or had a birth-weight below 1,500 g) and 1,032 controls (women who had infants without any of these characteristics) taken from the same age stratum as the cases. All pregnancies had occurred in 1980-81. Questionnaires were mailed to the women asking for information on their work situation, including questions about work with video screen equipment. Fifty-eight women were excluded for various reasons. Response rate was 93%--lower (89%) among women with spontaneous abortions than among women who gave birth (95%). As stress and smoking were associated with video screen work, the effect of video screen work was analyzed after stratification for stress and smoking--no statistically significant effect of video screen work was seen but odds ratios were above 1. Crude odds ratios for video screen work were significantly elevated and showed a dose-dependent effect. This finding is discussed from the point of view of biases in the study: selective non-responding, recall bias, geographical variability, and lack of information on women who had induced abortions. Using questionnaire data for exposure rates in the populations studied, an estimate of the effect greater than or equal to 10 hr weekly work with video screens on spontaneous abortion rate was made. The point estimate was 1.04 with a 95% confidence interval of 0.9-1.2. Analysis of 44 infants with birth defects whose mothers had worked more than 10 hr/week with video screen equipment compared to 30 infants with birth defects whose mothers had not used such equipment in early pregnancy showed no signs of specificity in the type of birth defect.