Based on attitude—behavior theory, it was hypothesized that computer use would enhance beliefs about self-perceived computer confidence, which would in turn affect attitudes towards computers. Primary level students (N = 723) completed self-report surveys that measured these three constructs. Covariance structural analyses revealed that (a) computer use positively affected computer confidence, and (b) computer confidence positively affected computer attitudes. Unexpectedly, direct computer use had a negative effect on computer attitudes, when confidence was held constant. Results suggest how computer educational environments might be improved.