We conducted a case-control study to investigate the relation between prenatal exposure to x-rays and childhood cancer, including leukemia, in over 32,000 twins born in Connecticut from 1930 to 1969. Twins as opposed to single births were chosen for study to reduce the likelihood of medical selection bias, since twins were often exposed to x-rays to diagnose the twin pregnancy or to determine fetal positioning before delivery and not because of medical conditions that may conceivably pre-dispose to cancer. Each of 31 incident cases of cancer, identified by linking the Connecticut twin and tumor registries, was matched with four twin controls according to sex, year of birth, and race. Records of hospitals, radiologists, and private physicians were searched for histories of x-ray exposure and other potentially important risk factors. Documented prenatal x-ray exposures were found for 39 per cent of the cases (12 of 31) and for 26 per cent of the controls (28 of 109). No other pregnancy, delivery, or maternal conditions were associated with cancer risk except low birth weight: 38 per cent of the cases as compared with 25 per cent of the controls weighed under 2.27 kg at birth. When birth weight was adjusted for, twins in whom leukemia or other childhood cancer developed were twice as likely to have been exposed to x-rays in utero as twins who were free of disease (relative risk, 2.4; 95 per cent confidence interval, 1.0 to 5.9). The results, though based on small numbers, provide further evidence that low-dose prenatal irradiation may increase the risk of childhood cancer.