The reliability and predictive validity of a parental screening instrument based on the Griffiths Mental Developmental Scale was investigated in a Swedish population of 3245 18-month-old children. Parental and professional assessments were compared. 2 per cent of low-scoring children were tested with the Griffiths Scale, and correlation with parental assessment was 0.87. The prevalence of mental retardation and of learning disabilities were investigated in follow-up studies at eight and 14 years, and these studies identified all mentally retarded children at those ages among the low-scoring or attrition cases at 18 months. Other learning disabilities were reported for 51.2 per cent of low-scoring children, compared with 18.5 per cent of controls. A comparison between parental assessment scores and test scores at 18 months with follow-up results showed that the two assessment methods yielded similar predictions.