This paper addresses the following questions: How large are the output costs of disinflation in Canada? Are these costs temporary, as predicted by natural-rate models, or are they permanent, as predicted by hysteresis models? Are the costs of disinflation higher at lower rates of inflation? Are they higher when the economy is at or below full employment? The answers to these questions are sought within the context of an expectations-augmented Phillips curve—a framework that permits direct calculation of the output-inflation trade-off. Our estimation results imply that the output loss of reducing inflation by 1 percentage point (the so-called sacrifice ratio) is around 2 per cent, which is lower than many other estimates. Moreover, we find no evidence of hysteresis in Canada, which means that this loss is temporary and not permanent. Finally, we could find no evidence that the slope of the Phillips curve is non-linear—the costs of disinflation do not appear to vary with either the level of inflation or the degree of excess supply.