Attempted to evaluate the contribution of rehearsals to 36 college students' performance on a continuous recognition task. Using 3-digit numbers as stimuli to minimize encoding variability, the spacing between successive item repetitions was systematically manipulated in variable rehearsal (VR) and constant rehearsal (CR) conditions. Performance on the 3rd presentation of critical items in both rehearsal conditions was observed following a constant long-term retention interval of 25 items. Results show that (a) the VR condition resulted in an inverted U-shaped lag function; (b) the CR condition, while attenuating any systematic effect of lag, produced a higher overall level of retention; and (c) rehearsals in both CR and VR conditions paralleled the performance data quite closely. These findings suggest that when contextual variability is reduced, performance on a recognition task may be mediated via rehearsal frequency. (20 ref) (PsycINFO Database Record (c) 2006 APA, all rights reserved)