The validity of the Student-Teacher Relationship Scale (STRS) was examined in a preschool community sample (N = 925) using confirmatory factor analysis (CFA). Factorial invariance across genders was also investigated as was concurrent and discriminant validity. Indicators of validity were teacher-rated social competence, problem behavior, and mental health, as well as the children's score on the Peabody Picture Vocabulary Test. Results showed that the original three-dimensional model of the STRS was not satisfactorily confirmed. However, a slightly modified 25 items, 3-factor version of the STRS showed an acceptable fit. Model fit was acceptable for both boys and girls. The modified three-factor model evidenced good concurrent validity. The discriminant validity of the dependency versus the conflict subscale was somewhat unsatisfactory, mainly due to imperfections in the dependency subscale. (Contains 1 figure and 6 tables.)