Antisaccade and prosaccade (PS) performance were studied in a large cohort of females (age range, 42-74). Antisaccade performance was assessed in two variants of the task, a "traditional" antisaccade (TA) task, in which no visual stimuli were present at the saccade goal, and a visually guided antisaccade (VGA) task, in which small visual stimuli were present at the possible saccade goals prior to the imperative visual stimulus. Directional error frequency was similar in the two antisaccade tasks. However, reaction time (RT) was ~33 msec longer in the VGA task than in the TA task. Across participants, average saccade amplitude of prosaccades and TAs were both correlated with those of VGAs, but not with each other. TAs had a hypermetria that increased with age. Saccade amplitude variability was much higher for TAs than for PSs and VGAs. Saccade polar angle variability was low for all 3 tasks. Age diminished performance with modest task dependence, except for an increase in TA hypermetria. These results suggest that the generation of antisaccade directional errors does not depend on visual target presence at the saccade goal, that antisaccade RT can be affected by target presence, that age can increase saccade hypermetria in the absence of visual guidance, and that visually guided antisaccades are governed by distinct voluntary and visually guided saccade mechanisms. Moreover, these results suggest that an understanding of human motor performance benefits from the use of a participant pool with a larger age range than is used in most studies.