A modified Swedish version of Buckley's Self-Evaluation Scale for psychotherapeutic skill was used in this study. An extensive sample of students from three universities was investigated. Two types of scale were used, Skill and Change in skill. Skill was rated by the supervisees both before and after supervision. Furthermore, Change after supervision was rated both by supervisees and supervisors. After Principal Component analyses, Varimax rotation, three dimensions (Containing emotionally loaded issues; Psychodynamic understanding; Mastery of working alliance) were interpreted in the supervisee ratings. However, the supervisors differentiated their ratings to a much higher degree than the supervisees. The factor structure of Skill, supervisee ratings before supervision, was used for calculating scales for all four rat-Marie-Louise Ögren, PhD, is Senior Lecturer, ing formats corresponding to the three factors. There were significant increases in Skill and positive Changes of skill after group supervision. The supervisors evaluated supervisees' Changes of skill higher than the students did themselves. Significant correlations of slightly more than medium size between supervisor and supervisee ratings of Change were obtained, and this was interpreted as confirming the validity of the supervisee ratings. Skill, in the dimension Mastery of working alliance, differed between groups depending on the group supervisors according to supervisee ratings after supervision.