The effectiveness of a computer-assisted balance training was compared with a home based exercise program in healthy elderly subjects. Twenty-four physically active community dwelling elderly (median age 71 years) participated in the study. Balance was assessed by a summary performance score, measuring postural stability while standing, reaching, stepping and walking. Subjects of the computer-assisted balance training group could considerably improve their performance on the training device (p = 0.0078) as well as in the summary balance score (p = 0.0176). No improvement of balance was documented in the home-based exercise group. A superiority of the computer-assisted balance training exercise protocol over the home-based exercise group could not be documented (p = 0.3934). The study indicates that a computer-assisted balance training, focussing on one motor skill, can improve balance. It is suggested that a home-based exercise program is ineffective to improve balance in elderly persons, if there is no direct supervision.