Mobility in older adults is associated with better quality of life. However, evidence suggests that older people spend less time out-of-home than younger adults. Traditional methods for assessing mobility have serious limitations. Wearable technologies provide the possibility of objectively assessing mobility over extended periods enabling better estimates of levels of mobility to be made and possible predictors to be explored. Eighty-six community dwelling older adults (mean age 79.8 years) had their mobility assessed for one week using GPS, accelerometry and self-report. Outcomes were: number of steps, time spent in dynamic outdoor activity, total distance travelled and total number of journeys made over the week. Assessments were also made of personal, cognitive, psychological, physical and social variables. Four regression models were calculated (one for each outcome). The models predicted 32 to 43% of the variance in levels of mobility. The ability to balance on one leg significantly predicted all four outcomes. In addition, cognitive ability predicted number of journeys made per week and time spent engaged in dynamic outdoor activity, and age significantly predicted total distance travelled. Overall estimates of mobility indicated step counts that were similar to those shown by previous research but distances travelled, measured by GPS, were lower. These findings suggest that mobility in this sample of older adults is predicted by the ability to balance on one leg. Possible interventions to improve out-of-home mobility could target balance. The fact that participants travelled shorter distances than those reported in previous studies is interesting since this high-functioning subgroup would be expected to demonstrate the highest levels.