While numerous residential programs for mentally retarded adolescent offenders have been established throughout this country, no empirically‐based population characteristics have been compiled which are sufficient to guide their systematic design. This study assessed 65 such young offenders along various treatment‐relevant dimensions, and used the resultant data to portray an average, hypothetical youth. Based on that profile, a compatible habilitation model was derived. It was concluded that a basic point economy, augmented by timeout procedures and supported by a secure (locked) setting, seems to be a viable approach for initiating treatment with these youth. Thereafter, programming should continue in a nonsecure group home, and finally, on an outpatient basis, in the generic community service system.