This study investigated the effects of imagery training on reactive agility and whether reacting to unpredictable stimuli could be improved using imagery. Forty-seven female athletes (M_age = 21.51, SD = 2.32) were randomly assigned to either a three-week physical training, imagery training, or control condition. Physical training condition involved physically rehearsing the reactive agility task, whereas the imagery training condition involved imagining the presenting stimulus and performing the reactive agility task. The control condition did no reactive agility training. A 3 (training conditions) x 7 (reactive agility performance components) mixed-model MANOVA was conducted to examine changes in reactive agility performance from the training interventions. Physical training improved decision time components and overall reactive agility performance. Imagery training improved Stimulus-Decision Time and Stimulus-Foot performance, but not overall reactive agility performance. No performance improvements occurred for the control condition. Findings support imagery use for the decision time variables associated with light-stimulus reactive agility performance. The lack of overall reactive performance improvement may indicate that imagery training is not effective for all components of perceptual-motor performance. Performance change inconsistencies appear to indicate that participants may not have generated unpredictable stimuli during imagery. Future investigation as to whether imagery improvements translate to sport-specific reactive tasks is needed.