A test of the effectiveness of the direct instruction model as an instructional method was conducted in Sydney (Australia) by administering pre- and post-tests to fifth and sixth year students taught computer literacy and computer programming skills in specially designed 5-week courses. Nine schools, some independent and some government-administered, participated in the study, which randomly assigned each of 31 classes comprising over 1,000 students to 1 of 4 instructional methods. Statistical analysis of the results indicated that the most highly structured direct instruction was the most effective. This method featured structured oral presentation of rules, class oral repetition of the rules, emphasis of the rules in written materials, varied response formats, close teacher supervision of written work, and provision of correction and reinforcement after each exercise. The least effective method involved presentation of examples by the teacher before the relevant rules were described, repetition of the rule by randomly selected individual students only, deemphasis of rules in written materials, a constant response format, no teacher supervision of written work, and inconsistent use of correction and reinforcement. A side benefit of the study was the effective exposure of hundreds of students to computers and computer use. (Author/PGD)