Modeling and physical guidance were compared for teaching 24 retarded males (mean age = 13.5, mean IQ = 25.3) to use a knife, fork, and spoon. Initial premeal training methods resulted in no improvement over baseline. The methods were revised to extend throughout meals, and a third method (physical guidance with praise) replaced a former control procedure. Methods involving physical guidance produced significant improvement while modeling did not. All groups then received physical guidance with praise which resulted in their achieving similar levels of performance. Follow-up observations revealed that correct utensil use continued under maintenance conditions.