This study investigates the Computer Attitude Scale (CAS) in terms of reliability, factorial validity, and fit to a unidimensional model. The sample for this study consisted of 265 teachers from 20 schools and school districts in one state who attended evening and weekend classes that emphasized teachers teaching teachers about classroom applications of computers and software. Participants were given a computer attitude scale as a pretest and as a posttest 9 months later; a workshop evaluation; and a self-evaluation of teaching skills including computer use. The results of this study suggest that: (1) the CAS is a reliable instrument for measuring teachers' attitudes toward and impressions of the educational applications of computers; (2) the CAS can be used as a total score or separated into four related subscales; (3) pretest and posttest differences can be used as one indication of the effect of training in use of computers in schools; and (4) a pretest can be used to identify potential training emphases regarding teachers' attitudes about computers. Two tables display the data on item characteristics for each factor and differences between item responses from pretest to posttest. (ALF)