We assessed the ability of procedural learning and gestalt cognition in patients with schizophrenia by push-button task and tree-drawing test, respectively. The study subjects were 30 patients with schizophrenia and 35 healthy individuals. The patients fulfilled the diagnostic criteria specified in the International Classification of Diseases, Tenth Revision and healthy individuals were all employees of private psychiatric institutions. We used a scale for measuring work ability and classified the patients into two groups of the same size on the basis of the measurements, and called them high-autonomy group and low-autonomy group. In the push-button task, we used an apparatus with 18 buttons whose lighting was controlled by computer, and the lighting order was determined to make the path of lighting form two types of gestalt. Complete push of 18 buttons defined a trial, and we assessed the ability of procedural learning by reduction in reaction time with trial. In the tree-drawing test, we classified the drawing into 3 types, i.e., positive, negative and normal, on the basis of the gestalt of the tree drawn. In each of the two types of gestalt and their combination, the reaction time at respective trials significantly (p<0.0001) decreased in the order of low-autonomy group, high-autonomy group and healthy group, and the reaction time in the respective 3 groups significantly (p<0.0001) decreased with repetition of trials. However, no significant difference was observed among 3 groups in the degree of decrease with trial (p>0.14). The frequency of positive drawing was highest in low-autonomy group and monotonically decreased in high-autonomy group and healthy group, while the reverse order held in the frequency of normal drawing, and the frequency of negative drawing was highest in high-autonomy group. The correlation between the group and the tree-drawing type was significant (p<0.0001). The present study suggests the usefulness of push-button task and tree-drawing test for assessing recovery process in schizophrenia.