This study was designed to investigate whether speech therapy enhances language recovery in severe aphasia and whether or not the method of speech therapy used affects recovery if it occurs. Thirty-one post CVA, severe expressive-receptive aphasics were assigned to one of three groups: programmed instruction, nonprogrammed instruction, and no treatment. All groups were comparable with respect to age, duration of symptoms, sex distribution, and education. Patients were initially tested to establish their level of functional language skills and ability to perform the terminal behaviors to be taught. Patients in the treatment groups received up to 40 hours of therapy. Posttests were administered individually after each terminal behavior was taught. All posttests were readministered at the termination of treatment and again one month after termination to determine the degree to which a behavior had been retained. There were no significant differences in outcome under the three treatment conditions. Results strongly suggest that current speech therapy does not modify verbal behavior in this population. A more realistic approach to the overall rehabilitation management of the severe aphasic, and more careful selection of patients who will receive speech therapy is suggested.