Recent research on e-learning shows that blended learning is more effective than face-to-face learning. However, a clear empirical response has not been given to the cause of such improvement. Using a data set of 9044 students at two Catalan universities and a quasi-experimental approach, two possible hypotheses identified in previous research are studied. The results show that the principal cause of the improvement is not, in itself, the increase in time spent online for educational purposes. Rather, increasing the time devoted to studying online is only useful when it takes place as some form of interactive learning. The educational implications of these results are discussed.