This research proposes and tests a reactivation theory of spacing effects which assumes them to be caused by the reactivation of representations whose activation levels have decayed within spaces, and recall rates to increase as the amounts of the reactivation increase. According to this theory, therefore, recall rates are supposed to increase sharply when the spaces are small enough for the representations formed at the first presentation to remain in working memory because their activation would decay rapidly and, on the contrary, the amount of reactivation at the second presentation would increase rapidly. They also are supposed to be rather high and stable when the spaces are large enough for the representations to be transferred to and consolidate in long-term memory because their activation levels there would be relatively low and stable and the amount of reactivation would be relatively high and stable. Two experiments and two simulation were conducted to test this theory. In Experiment 1, recall rates at various spaces and presentation times were examined, using two-digit numbers for stimuli on which encoding variability should have little effect and the effects of spaces themselves should be clear, and the results were consistent with most of the suppositions above. In Simulation 1, a simulation model of the reactivation theory was made on the experimental data and a simulation was conducted to find the estimated recall rates approximated to the measured recall rates in Experiment 1. In Simulation 2, recall rates on various expanding space conditions in three-times presentation were predicted. Then in Experiment 2, recall rates on the same conditions with those in Simulation 2 were measured to find they are very close to the predictions in Simulation 2. Finally, this theory was applied to explain several previous inconsistent experimental results found under the other theories of spacing effects.