Undergraduate students (N = 83) were pretested on four standard tests of visual-spatial skill. Half were given 3 hours of training relevant to the spatial tasks presented by three of the tests. All subjects were then posttested. The hypothesis that spatial "ability" is susceptible to practice and training effects was strongly supported. Multivariate analysis of variance showed that experimental subjects improved significantly more than control subjects, male and females improved equally and substantially, and training effects generalized to an untrained spatial task. The hypothesis that females score lower on spatial tests because they lack relevant practice was also supported; when female experimental subjects were compared with male control subjects on the posttests, the sex-related pretest difference favoring males was eliminated.