Cases of nonspecific mental retardation (MR) born in British Columbia between 1952 and 1970 ascertained through the British Columbia Health Surveillance Registry were linked by birth registration number to family sibships from computer-linked groupings of birth and marriage records in British Columbia. It was possible to retrieve family information for 97% of the cases by this method. Because of good ascertainment and relatively large sample size, the 1952-1965 birth cohort comprising 2,209 index cases was selected for calculations of overall risks and recurrence risks to sibs categorized by sex, MR level, associated neurological disability, and singleton versus multiple birth. The overall risk of affected individuals among all sibs was 4.4 +/- 0.6%, which was about ten times greater than the minimum population incidence of nonspecific MR. The risk among subsequent sibs of the first affected case in a family was 3.7 +/- 0.8%. These risks varied depending on sex, MR level, and whether the mental retardation was associated with hydrocephalus, microcephalus, cerebral palsy, or epilepsy. The recurrence risk after two affected individuals was 12 +/- 7%--about three times greater than after one affected individual. Even though the frequency of MR is greater among twins than in the overall population, the recurrence risk of nonspecific MR was not significantly different for index cases from either singleton or multiple births.