Physicians (N = 803) were contacted via postal survey and given two sets of efficacy measures for drug treatments in Alzheimer's disease: (a) the time that patients spend in a mild or moderate state of disease; (b) levels of modification to disease progression in the areas of cognition, behaviour, and mood, and ability to perform basic activities of daily living. Physicians reported that they would prescribe a hypothetical, new Alzheimer's disease medication if it would allow patients to remain in their current disease state for 15 (mild) or 11 (moderate) additional months. Most physicians required a permanent halt to, or some reversal of, disease progression as a prerequisite for prescribing; a few required substantial reversal. More stringent efficacy requirements were negatively associated with physicians' current prescribing of cholinesterase inhibitors to persons with Alzheimer's disease, although the effects were either small (odds ratio = 0.99) or not statistically significant at the 5 per cent level. The results suggest that physicians with stringent efficacy requirements for clinically relevant efficacy measures are less likely to prescribe cholinesterase inhibitors.