Two experiments were carried out to examine the relationship between sensory and/or semantic variability of stimulus words and the effect of massed versus distributed repetition on free recall. In the first experiment stimulus words were presented orally four times, each presentation by a different experimenter's voice. The subject's attention to words was measured in terms of reaction time to subsidiary signal tones. The variability of voices facilitated the attention to words in the massed condition but did not enhance recall performance. In the second experiment stimulus words presented visually were accompanied by constant or varied encoding sentences presented auditorily to induce variable semantic contexts. The variability of encoding sentences increased recall performance in the massed condition, but the influence of distributed repetition was not completely eliminated. These results suggest that the spacing effect in free recall is related to variability in semantic encoding but not to sensory or attentional components.