A comparison was made between an interspersal and a concurrent procedure in teaching picture names to three mentally handicapped children. During the interspersal procedure a picture thats name was being trained was alternated with pictures already known; during the concurrent procedure a picture thats name was being trained was alternated with other pictures thats names were unknown. An ABA design with counterbalancing (BAB) was used. The children learned naming responses more rapidly when trained by the interspersal procedure than by the concurrent procedure. Weekly retention tests on pictures learned to criterion during the week showed no consistent difference between the two procedures in percentage of learned picture names retained. Weekly generalization tests showed that picture names that were retained in both conditions tended to generalize equally to a different setting and tester, and to the objects depicted in the pictures.