The relationship between extreme temperament in infancy and clinical status at 4.7 years of age was studied in temperamentally different groups of infants matched for sex and SES, and subselected from a large birth cohort representative of the general population. The effects of certain dimensions of family functioning and of other risk factors were examined. By itself, extremely difficult temperament in infancy had no strong direct association with clinical outcome at four years of age, whereas temperament assessed at four, family attitudes to discipline, and stressful events did. However, extreme temperament in infancy might indirectly affect outcome through its association with temperament at four. The interplay between adverse temperament and parental attitudes of discipline previously observed in middle childhood might have antecedents in pre-school years.