The purpose of this study was to determine the effects of rating purpose and anonymity on the psychometric characteristics of self-ratings. Employees of the United States Marshals Service rated their proficiency on 66 job-related knowledges and skills, and on overall performance. Instructions stated. that the purpose of the ratings was either administration, research, i.e. training needs assessment. In addition, subjects were either required to sign their names or to remain anonymous. Each employee's immediate supervisor also provided ratings for each subject. Self-rating leniency did not vary as a function of anonymity. Self-ratings made for training purposes were less lenient than ratings made for administration or research. Self-ratings demonstrated less halo than supervisor ratings. In addition, self-ratings exhibited less variance and more leniency than supervisor-ratings. It is suggested that self-ratings may be a useful method of identifying training needs when used in conjunction with supervisory evaluations. / M.S.