The Common Beliefs Survey-III (CBS-III), a factored measure of general irrational beliefs, has satisfactory convergent and discriminant validity. The Situational Self-Statement and Affective State Inventory (SSSASI), also with acceptable basic psychometric properties, assesses specific thoughts and feelings in response to vignettes describing frustrating events. Both inventories are potentially useful in assessing dimensions important to REBT, but information on test-retest reliability is needed. The CBS-III and the SSSASI were administered to 101 undergraduate students on two occasions, two or three weeks apart. Satisfactory test-retest reliability coefficients were obtained on all subscales of both inventories, but in the case of the SSSASI reliabilities were generally stronger when the same form of the test was used on both occasions (Form 1 presented the thoughts before the feelings, whereas Form 2 presented the feelings before the thoughts). Further exploration showed that responding to the thoughts before the feelings significantly attenuated scores on the feelings, but responding to the feelings before the thoughts had no effect on scores on the thoughts. This unexpected finding also held true in a previously unexamined data set obtained with a different sample of students several years earlier. Taking careful stock of one's emotional reactions before embarking on cognitive modification interventions seems essential to REBT, and it follows that asking questionnaire respondents to rate their affective states before they rate their self-statements could be desirable conceptually. These results provide empirical support for that strategy.