Currently, treatments for amblyopia are occlusion or penalization of the stronger fellow eye. Fewer than 30% of patients improve stereoacuity using these treatments. In order to improve these outcomes, this group proposes a treatment to stimulate the stereoacuity through perceptual learning in a game format for use at home. The aim of this study was to determine whether perceptual learning with random dot stimuli (RDS) in the form of a computer video game improves stereopsis in patients with a history of amblyopia. Thirty-two stereo-deficient patients (7 to 14 years old) previously treated for amblyopia participated in a prospective, randomized, double-blind study. Participants followed a perceptual learning program at home using RDS software. In the experimental group, the demand of stereopsis was increased, until reaching the lowest detectable disparity. In the comparison group, the stimulation interval was a constant (840 to 750"). Stereoacuity was evaluated with the Randot Preschool Stereoacuity Test (RPST) and the Wirt Circles. Median compliance was 100% (interquartile range [IQR] = 78.50 to 100). Log10 stereoacuity outcomes were significantly different between groups (RPST, P = .041; Wirt Circles Test, P = .009). Median stereoacuity improvement with RPST was 50% (IQR = 0.00 to 75%) and 0% (IQR = 0.00 to 7.5%), respectively, for experimental and comparison groups (P = .008). Wirt Circles improvement was 46.42% (IQR = 6.25 to 73.75%) and 0% (IQR = 0.00 to 57.50%), respectively, for experimental and comparison groups (P = .089). Stereoacuity improvement was not different between groups when success was considered a 70% gain in RPST (P = .113); it was statistically different when success was considered a gain of two levels on Wirt Circles and stereoacuity 140" or less (P = .023). Stereoacuity remained stable after 6 months when measured with RPST, whereas it worsened in two subjects when measured with Wirt Circles. Direct stimulation of stereopsis at home using RDS in a game environment improves the stereoacuity in stereo-deficient subjects with a history of amblyopia.