PURPOSE: There are well-documented variations in the practice of medicine, so much so that it is not uncommon that for some doctors in the same clinical situations to undertake a “wait-and-see” policy, while others order a diagnostic test, and still others initiate treatment without testing. Any of these strategies may in retrospect prove to be wrong with realization that another alternative would have been preferable. This knowledge may bring a sense of loss, or regret. We asked if variation in practice could be explained by different decision-makers having a different amount of regret in case decision proved to be an erroneous one. METHODS: We extended our previously published model of acceptable regret to a generic three-choice decision-making problem (do nothing vs. test vs. treat). We defined acceptable regret as the one when making a wrong decision will not be particularly burdensome to the decision-maker. Taking into account acceptable regret (Ro), sensitivity (S), specificity (Sp), risks of tests (Tr), net benefits (B) and harms (H) of treatments, we derived three threshold equations that define a normative framework for the three-choice medical dilemma. RESULTS: We showed that if Ro<=(1-Sp)*H<(1-S)*B we could never order a test without exceeding acceptable regret. However, if Ro=>(1-S)*B>(1-Sp)*H we can always order a test without fear of exceeding acceptable regret. Since in most clinical situations (1-S)*B>(1-Sp)*H, this may explain why physicians tend to overtest. Furthermore, we demonstrated that in the same clinical situation different decision-makers might decide to set their acceptable regret thresholds for withholding treatment to differ from the one for ordering a diagnostic test or for administering treatment. This in turn means that a normative model for one decision-maker may indicate that the most rational strategy is to “do nothing” while for the other it may point to ordering a diagnostic test and yet for the third committing to treatment may be the most rational course of action! CONCLUSIONS: Taking into account acceptable regret in decision-making can explain both practice variation and overtesting in today's practice of medicine.