The drug treatment of heart failure, once simple, has become complex. Apart from a loop diuretic and digoxin, most patients should now be receiving an angiotensin-converting enzyme inhibitor (or angiotensin II receptor blocker), a beta-blocker and spironolactone. Newer drugs, such as endothelin-receptor antagonists and combined blockers of converting-enzyme and neutral endopeptidase, might soon become available. When to introduce these drugs and what dose is optimal for any individual, are questions that currently vex clinicians. We proposed that plasma levels of the cardiac hormone brain natriuretic peptide (BNP, or better, its 1-76 amino-acid N-terminal fragment, N-BNP), would provide an objective index for guiding drug treatment in patients with established, stable cardiac failure. In a pilot study, 69 patients were randomized to drug treatment based on clinical criteria, or based on plasma levels of N-BNP. After a median follow-up of 9.6 months, those in the N-BNP group had fewer clinical end-points than those in the group managed by clinical criteria alone (19 vs 54; P= 0.02). These preliminary data encourage the concept that the increasingly complex pharmacotherapy for heart failure, both chronic (as in this trial) and acute, might best be guided by an objective measure such as plasma levels of BNP or N-BNP.