A total of 40 counselors, ranging from novice trainees to experienced professionals, conducted 25-min counseling interviews with a female client, and then completed instruments measuring frequency of task facilitative and task distractive self-talk and quality of clinical hypothesis formulation. It was hypothesized that these measures of counselor internal dialogue would be predictive of counselor performance (clients' and trained raters' evaluations) once the effects of counselor gender, education level, and years of experience were accounted for. Cognitive variables were assessed, using standard multiple regression techniques, for their combined and unique contributions to the overall prediction equation. The major finding of the study was a positive relationship between higher quality clinical hypothesis formulation and higher levels of facilitative performance during counseling sessions.Counselor training research has increasingly emphasized cognitive processes during the past decade. This trend is reflective of a more global movement toward cognitive approaches in the behavioral sciences (Mahoney, 1974;Meichenbaum, 1977) and is viewed by many counselor educators and researchers as a positive stimulus for moving toward more comprehensive training models (Fuqua, Johnson, Anderson, & Newman, 1984; Kurpius, Benjamin, & Morran, 4985;Schmidt, 1979;Stone, 1980). They contend that traditional models have narrowly focused on behavioral skills and must be expanded to account for complex interrelationships between counselor trainee cognitive and behavioral processes.Within the counselor education field, cognitive-behavioral studies initially focused upon the relationship of trainee cognitive attributes to counseling performance. Such attributes as conceptual level (Goldberg, 1974;Holloway & Wolleat, 1980), cognitive style (Handley, 1982), and cognitive complexity (Lichtenberg & Heck, 1979) were found to be associated with various measures of counselor effectiveness. Studies of this nature have, however, been criticized because they assess relatively stable dimensions of cognition and reveal little about what actually occurs during the counseling session. Martin (1984), for example, has called for future investigations to obtain more direct data about specific, and presumably more changeable, cognitive operations in ecologically valid counseling contexts.Recent counselor training developments have begun to shift toward a focus on more specific and changeable aspects of counselor cognition. Based on such procedures as self-control (Goldfried & Merbaum, 1973), self-instruction (Meichenbaum, 1977), and mental imagery (Singer, 1974), I would like to thank Greg Brack for the great deal of help he provided on computer work. I would also like to thank Patti Parrett, Cathy Riggs, and Joyce Stout for their many hours of volunteered service in organizing and carrying out this study.