A bivalent counterbalanced within-subjects research design was used to determine the effectiveness of vocabulary instruction across two instructional conditions with 30 3 to 4-year-old at-risk preschool children. Instruction presented vocabulary words via static pictures with one subgroup and via projection and animation with the other. Conditions then were reversed. Dependent variables were children's serial free recall ability in specified categories and ability to name aloud selected pictures. There was a pretest, instructional, and posttest phase for each condition. Maintenance data were collected 12 weeks after the last posttest session. Results indicated no difference between conditions. Posttest and maintenance data indicated an increase in serial free recall and picture naming of targeted vocabulary that was maintained above the pretest level for all dependent variables. (Contains 1 table.)