The design of a school anxiety questionnaire is described. The model predicts a maximal relationship between anxiety and performance when type of anxiety potential, type of stress, and type of behavior are in parallel. The test, developed with intermediate school children, is a five scale 105 item multiple choice questionnaire with responses on a 5-point Likert scale. It is administered to children in their classroom with the teacher absent. Instructions are read aloud at 5 second intervals by a tape recorder. The student marks his response on a mark sense sheet for data processing. Three hundred and twenty student responses were factor analyzed using a principal axis factor with a normalized verimax rotation. To examine factor structure stability over age, data for third and sixth grade subjects were factored separately and their rotated structures compared using Kaiser's coefficient of factor stability. In spite of their brevity the SAQ Anxiety scales reflect adequate levels of reliability and have predictive validities that are in theoretically meaningful directions, are of a magnitude not commonly reported, and are generally replicable, both across age groups and across studies. (DJ)