This study first investigated the possibility of employing mildly mentally retarded residents to observe the mealtime behaviors of lower level mentally retarded residents. It was found that the residents could rate mealtime behaviors reliably (mean reliability = 95%, range: 72% to 100%). The residents were then employed as reinforcing agents, to determine if a retarded "trainer" could modify the observed mealtime behaviors within a multiple baseline design. Since an experimental effect was demonstrated on four (proper use of napkin, clean face, clean table, and proper use of utensils) of the five target behaviors, it was concluded that such change agents could modify the mealtime behavior of lower functioning residents. EXPERIMENT 1 SubjectsTwo mildly retarded residents of one living unit (Wechsler Adult Intelligence Scale IQ 62