Abstract Fifteen persons with profound mental retardation were divided into two groups. One group was identified with chronic training needs by habilitative staff and the other group served as a control. In an attempt to identify a reinforcer, each participant received a preference assessment and a simple, low-effort treatment procedure. In Experiment 1, only individuals who approached at least one stimulus on 80% or more of the preference assessment trials (“high preference”) showed reinforcement effects in treatment. However, three individuals showing high preference failed to show treatment effects. All persons identified with chronic training needs failed to show reinforcement effects. Experiment 2 analyzed characteristics of the two groups and found significant differences in overall movement and response latency. Limitations of the current reinforcement technology were apparent for identifying reinforcers in the group with chronic training problems. Research is suggested for evaluating training alternatives for people with profound multiple disabilities who move very little or who respond with very long latencies.