A statistical measure, the detection level, that conveys information about the risk of type II errors in a way that is easily understandable to the statistically unsophisticated decision maker is proposed. The detection level answers the question: "How big should an effect have been in order to be detected in this way?" Given a statistical test with the critical level (size) alpha*, the detection level of this test is the smallest effect that would, with at least probability 1-alpha*, lead to a significant outcome (P < alpha*) of the test. It is proposed that detection levels should be calculated as a standard procedure in toxicity reviews, and in particular in reports from toxicologists to decision makers.