The “reasonable apprehension of bias” test for judicial disqualification has been a fixture of Canadian law for many years, despite a considerable amount of litigation concerning judicial impartiality. The test itself has remained fundamentally unaltered and is well accepted in the jurisprudence. Unfortunately, the application of the test continues to generate difficulties for judges who need to use it to make decisions in marginal cases. Based on previously published doctrinal and empirical research, our goal in the present paper is to suggest modifications to the test that will better explain the existing jurisprudence and make it easier for judges to understand when recusal is or is not necessary in marginal cases. We begin considering the advantages of and suggest that in order to be useful, any refinement to the test must to the greatest extent possible preserve those advantages. In the second part of the paper we consider why inconsistent application of the test in marginal cases is a concern. This is followed by a more detailed consideration of the ways in which the existing test, and the jurisprudence explaining and applying it, are problematic. The fourth part of the paper proposes a modification to the “reasonable apprehension of bias” test that is designed to address these shortcomings while preserving the key advantages of the existing test.