The purpose of this study was to examine the value of various durations of ambulatory ECG recording with regard to providing useful prognostic information. The authors explored a decision theoretic approach to determine the most useful period of monitoring for making a treatment decision based upon postulated benefit-to-risk ratios of antiarrhythmic therapies. They used data collected as part of the Beta-Blocker Heart Attack Trial (BHAT), a randomized clinical trial of propranolol versus placebo in 3,837 post-myocardial-infarction patients. In BHAT, 1,336 placebo-treated patients had a 24-hour ambulatory ECG that had at least 23 readable hours. Sensitivity and specificity were calculated for eight definitions of ventricular arrhythmia using either total mortality or sudden death (death within one hour of symptoms) as an endpoint. These indices were obtained using the first 1, 2, 4, 6, 12, and 24 hours plus a random hour, a random daytime hour, and a random nighttime hour of the 24-hour ECG of 1,336 placebo-treated patients. The study showed that in the case of high-risk, low-benefit therapies, no test is needed to make a treatment decision. No one should be treated. In the case of high-benefit, low-risk therapies, again, no test is required. Everyone should be treated. For therapies in the middle benefit-to-risk ratio range the most appropriate test for a treatment decision changes from the very specific to the most sensitive. Twenty-four hours of ambulatory monitoring is usually not necessary for a treatment decision, since four hours is likely to be sufficient.