Variability of practice has received much attention in the motor learning literature. The variability of practice hypothesis states that rather than remembering specific sensory consequences, learners use feedback to establish certain rules or schema. These schema enable the learner to establish a generalized motor program. Stability training is used for a variety of reasons (e.g. rehabilitation, reducing falls, enhancing sports performance). The Biodex Stability System (BSS) is an instrumented circular tilt board that moves in a 360° horizontal plane. Changing the resistance force applied to the platform can vary the platform stability. The degrees of tilt are measured and a stability index is calculated based on the deviation from horizontal. This index was used as the dependent variable in this study. While the reliability of the BSS has been established, we were unable to find research that examined the efficacy of different practice schedules on balance. The purpose of this study was to compare the effects of random, serial and constant practice on performing a dynamic balance task. Three groups of ten subjects performed the stability task on three consecutive days and completed nine 20-s trials each day. Stability levels 4, 6, and 8 were used, with level 8 being the most stable. Group 1 performed three trials at level 4, three trials at level 6 and three trials at level 8. Group 2 performed three trials at each level in random order. Group 3 performed one trial at each level (4-6-8) in serial order, repeated three times. Two days after the completion of the practice trials all subjects including a control group (no practice) performed a transfer task at stability level 2. A Mixed Model ANOVA showed non-significant group main effects (p>.05), but significant trial main effects (F=30.07, p=.000). All groups improved with practice across blocks of trials (3 blocks, 9 trials each) (p<.05). For the transfer task, significant group main effects (F=5.16, p=.002), but no trial main effects (3 blocks, 2 trials each) (p>.05) were found. Post Hoc results showed the random practice group was superior to the other groups (p<.05), whereas the block and serial practice groups were only superior to the control (p<.05). These results infer that despite no superior balance performance of any group during balance acquisition, random practice produced better results on the transfer task. Keyword(s): performance, physical activity, research