In order to identify differences in determinants of supportive and class use of computers, path modelling was applied in a sample of 468 primary school teachers. Independent variables were categorised in three levels: demographics (age and gender), computer experience (computer training, computer experience expressed over time, intensity of computer use), and attitude measures (general computer attitudes, attitudes toward computers in education, and technological innovativeness). Supportive and class use of computers are not related to the same set of variables. Supportive computer use was mainly predicted by computer experience variables and general computer attitudes. Strongest predictors of class use were technological innovativeness and gender. Yet, the degree of explained variance for class use of computer was considerably lower compared to supportive computer use. These results indicate the limitations of explaining complex forms of professional computer use on the basis of both individual determinants and quantitative models. The article concludes with some practical implications and recommendations for further research. (Contains 2 figures and 4 tables.)