Studies suggest that cognitive training and physical activity can improve age-related deficits in dual-task performances. However, both of these interventions have never been compared in the same study. This article investigates the improvement in dual-task performance in 2 types of exercise training groups and a cognitive training group and explores if there are specific dual-task components that are more sensitive or more likely to improve following each type of training. Seventy-eight healthy inactive participants older than the age of 60 (M = 69.98, SD = 5.56) were randomized to one of three 12-week training programs: aerobic training (AET) = 26, gross motor abilities (GMA) = 27, and cognition (COG) = 25. Before and after the training program, the participants underwent physical fitness tests, and cognitive evaluations involving a computerized cognitive dual task. The AET consisted of high- and low-intensity aerobic training, the GMA of full-body exercises focusing on agility, balance, coordination, and stretching, and the COG of tablet-based exercises focusing on executive functions. Repeated-measures analysis of variance on reaction time data revealed a group × time interaction (F(2,75) = 11.91, p < .01) with COG having the greatest improvement, followed by a significant improvement in the GMA group. Secondary analysis revealed the COG to also improve the intraindividual variability in reaction time (F(1,24) = 8.62, p < .01), while the GMA improved the dual-task cost (F(1,26) = 12.74, p < .01). The results show that physical and cognitive training can help enhance dual-task performance by improving different aspects of the task, suggesting that different mechanisms are in play.