‘Noise’ is the unjustified and unwanted variance in a set of judgments over comparable issues. Together with bias, Noise is a driver of error in decision-making. As argued by the authors of the bestseller ‘Noise: A Flaw in Human Judgment’, every set of judgments or decisions (in legal proceedings or otherwise) evidence statistical ‘Noise’, and more of it than is commonly believed. Such variance has corrosive, if often concealed, consequences in terms of fairness, efficiency and legitimacy. In this article we demonstrate that there is likely to be substantial Noise in international arbitration proceedings, which is driven by features inherent to the arbitral process (though further features also help mitigate it). We present our Noise Audit and identify examples of Noise in publicly-available awards. We conclude with a number of recommendations to minimize Noise, in order to forge a pathway towards greater consistency in international arbitration.