A program evaluation was undertaken at Southern Illinois University (Carbondale) to assess the broad, measurable effects of using computers to teach introductory college composition. Twenty-four classes were studied--twelve control classes and twelve experimental--with the experimental computer classes meeting in the lab for half of their instructional time. Data on the success of the program were collected from a range of sources: pre- and posttests of student writing under both impromptu and take-home conditions; pre- and posttests of writing anxiety; records on attendance, tardiness, withdrawals, and homework and essay assignment completion; end-of-term course evaluation by both teachers and students; and self-report data collected from teacher meetings and teacher logs. Results favored the use of computers, with computer students revising and improving their posttest essays (especially discourse-level features) at levels that are significantly better than regular students. Those students in experimental sections who chose to compose on computers at the end of the term outperformed the group as a whole and performed significantly better than those experimental students who chose to compose with pen and paper. Attitudinal data from both students and teachers also favored the use of computers. (Seven tables of statistical data are included and 20 references are appended, as well as handouts giving topics for pre- and posttests). (Author/ARH)