Two experiments used procedures similar to those used by R. L. Greene (1989) to test the 2-process theory of the spacing effect and, in particular, the contextual-variability subtheory that applies to free-recall performance. Experiment 1 obtained a spacing effect in free recall following intentional learning but not following incidental learning, contrary to a previous result supporting the 2-process theory. Experiment 2 replicated the incidental-learning results when a slow presentation rate was used. However, with a faster presentation rate, a spacing effect was obtained, and performance exceeded that of the slow-presentation-rate condition at the longest lag. Neither the contextual-variability subtheory of 2-process theory nor an alternative deficient-processing hypothesis was able to account for all of the data.