We report on an epidemiological-clinical study of the New York (NY) Longitudinal Study temperament model in a consecutive sample of children (N = 814) referred to a child psychiatric center. Temperament comparisons in this clinical population were made by using temperament normative values obtained in previous random samples of the general population in the greater Quebec City (Canada) area. Different clinical diagnostic groups (externalized disorders, developmental delays, and mixed disorders) were derived from a review of the entire hospital charts in which the interrater reliability was tested and performed "blind" to temperament scores. The diagnostic groups were confirmed through discriminant function analyses. The results (1) replicated, in this child psychiatric population, two factors of temperament similar to those previously found in random samples of our general population; (2) showed, in the psychiatric population of children, an overproportion of difficult temperaments on both factors; (3) confirmed conversely that a large proportion of children referred for a disorder did not present with an extreme temperament, and, therefore, an extreme temperament and a clinical disorder were not equivalent; and (4) suggested a specificity in the relationship between particular temperament factors and the type of clinical problem. Temperament factor 1 (withdrawal from new stimuli, low adaptability, high intensity, and negative mood) was found to be more associated with externalized disorders (opposition, conduct, or attention-deficit disorders), whereas temperament factor 2 (low persistence, high sensory threshold, and high mobility) was found to be more associated with specific developmental delays. The findings provided leads for future clinical research on temperament, family functioning, and child psychiatric diagnoses.