The following practical issues in psychopharmacotherapy of patients with mental retardation (MR) or other developmental disability (DD) are discussed, with some theoretical speculations: Persons with MR/DD have high rates of psychiatric disorders/symptoms, many of which are drug-responsive and not satisfactorily treated with other modalities alone. However, diagnosis is often complicated by concurrent multiple disorders (both psychiatric and medical/surgical), masking or distortion of symptoms, and communication impairments. Comorbidity and associated treatments should be considered in medication choice. Compared to psychiatric patients of normal IQ, patients with MR may be more sensitive to side effects and toxicity as well as responsive to lower doses, possibly related to less neuronal substrate, qualitative brain differences, or developmental stage. Unexpected or disappointing drug responses may also be related to such statistical quirks as "end-of-curve" phenomena. Ripples, ratcheting, and other ecobehavioral considerations influence treatment outcome and drug management. Target symptoms and expected drug benefits should be defined in consultation with caregivers as well as, when possible, the patient. Because drug responses of patients with MR are especially unpredictable, unreliable, sensitive to dose, and fraught with side effects, each medication trial might ideally be approached as a single-subject experiment on a compassionate protocol: quantify the baseline with ratings, behavior counts or other objective measures; start low and titrate slowly; monitor progress with repeated objective measures.