The causes of most childhood cancer remain elusive; some children clearly have a genetic predisposition, but in the majority the relative contributions of environmental and host factors are not established. One approach to this question is through twin concordance studies, but only the most common malignancy, acute leukemia, has been studied to date, owing to the rarity of other forms of childhood cancer. The aim of the study was to determine the concordance rates for childhood cancer in twins, in order to clarify the importance of constitutional predisposition for a range of tumor types. Twins with cancer were ascertained through three cooperative clinical trials groups, a cancer-twin registry, and a large pediatric hospital. Subjects were sent a postal questionnaire requesting information on cancer concordance and zygosity. Data were obtained on 556 twins with cancer. Three twin pairs, out of 197 twin pairs (76 monozygous, MZ, twin pairs), were concordant for leukemia, giving an MZ case-wise concordance rate (5%) that is substantially lower than previously reported. The case-wise concordance for non-retinoblastoma solid tumors was 2.2%: Two twin pairs were concordant for CNS tumors, one was concordant for neuroblastoma, and two twin pairs were concordant for cancer but not for the type of cancer. The results of the present study, together with previous data from population studies of siblings and offspring, suggest that there is not in general a strong constitutional genetic component for childhood cancers other than retinoblastoma.