The objectives of this study were to verify the influence of training through generalized and supervised physical activities, on the general agility (GA) and upper-limb agility (ULA) of elderly women, in addition to verifying whether a relationship exists between these types of agility. Sixty women (59.7 ±5.9 years) were divided into two groups: a) trained group (TG) - participants of a supervised program of generalized physical activities, for at least one year, three weekly sessions of one hour each; b) an untrained group (UG) - with no practice of regular and supervised physical activities, besides every-day life activities. For GA evaluation, the agility and dynamic balance test designated by AAHPERD were applied and, for ULA, the plate tapping test designed by EUROFIT. The TG mean AAHPERD test time was 19.9 ± 2.7 seconds, and their EUROFIT Test TG score was 149 ± 23.2, while the UG achieved 21.7 ± 3.4 seconds on the AAHPERD test and scored 157 ± 24 points on the EUROFIT test. The difference between TG and UG was statistically significant for the GA test, but the same was not true of the ULA test. The correlation between GA and ULA results was r=0.51. It is concluded that for elderly women: a) regular practice of general physical activities under supervision improves general agility level, but not upper-limb agility; b) general agility scores are poor predictors of upper-limb agility.