Contrary to general findings in the attention and memory literature, some studies have shown that performing a secondary cognitive task produces an improvement in balance performance. The purpose of the present experiment was to investigate under what condition such an improvement would occur. Young and older adults were asked to hold as still as possible on a platform that measured sway while performing or not performing the encoding phase of the Brooks' (1967) spatial or non-spatial memory task. The difficulty of maintaining balance was manipulated by varying the availability of visual input and sway-referenced motion of the platform. Sway scores were computed based on the distance between the individual pressure centres and the average centre of pressure during each 20-s trial. The results indicated that both the spatial and non-spatial memory tasks improved balance for older adults under the most difficult balance condition.