Two experiments are reported in which patients who resided on continuing care psychogeriatric wards were exposed to an operant conditioning procedure. In the first experiment, the subjects were three female patients, all of whom were suffering from severe dementia. For two of the subjects, extended acquisition training was required before evidence of learning was found. Responding under fixed interval (FI) schedules of three different durations was well maintained by the third subject. Evidence of temporal control was found. The second experiment was a partial replication of the first. The subjects were two male patients who were suffering from mild to moderate degrees of dementia. They were exposed to FI schedules of three different durations. Responding was maintained for the 16 sessions of the study. Procedural modifications as well as some broader implications of these results are noted.