prevents symptomatic deterioration. So, a clinician seeking advice finds it in our guideline, but our recommendation should not be considered dictum. Our recommendation against routine warfarin anticoagulation in the absence of atrial fi¬ brillation is equally pertinent. There is no hard evidence for clinical benefit, and the incidence of systemic emboli is small. But a knowledgeable clinician may calculate that the poten¬ tially catastrophic nature of systemic emboli and the minimal risk of hemorrhagic complications offered by current antico¬ agulant recommendations argues for anticoagulation. Thus, although we do not recommend routine anticoagulation, I would not prohibit it.Other examples of recommendations that should not be considered dictum include those regarding ß-blockers, hydralazine, and isosorbide dinitrate, ACE inhibitor use with¬ out diuretics, and revascularization. Each of these areas re¬ quires sophisticated interpretation and judicious individual application.Which recommendations might be considered dictum and enforced by payer plans and government agencies? I propose the following criteria: (1) the recommendation is based on either (nearly) universally accepted clinical trial data or (nearly) universal sound practice; (2) the recommendation is (nearly) globally applicable; and (3) the recommendation is readily testable through reasonable documentation. Examples include measurement of left ventricular function in patients with clinical heart failure, ACE inhibitor use in patients with a low ejection fraction, unless contraindicated, and diuretic use in patients with volume overload.Clinical practice guidelines are invaluable as consensus statements, consolidating information and offering guidance to clinicians who must prudently interpret and apply their recommendations. The challenges are keeping guidelines au¬ thoritative and current, disseminating them widely, and re¬ sisting temptation to elevate guidelines above their calling, ie, avoiding efforts to substitute them for sound clinical judgment.