This paper shows that if the Bank of Canada is optimally adjusting its monetary policy instrument in response to inflation indicators to target 2 per cent inflation at a two-year horizon, then deviations of inflation from 2 per cent represent the Bank's forecast errors, and should be uncorrelated with its information set, which includes two-year lagged values of the instrument and the indicators. Positive or negative correlations are evidence of systematic errors in monetary policy. The econometric evidence suggests that, over the past decade, the Bank has, on average, responded optimally to indicators of inflation, being neither too aggressive nor too timid in raising or cutting the overnight rate. While responding optimally on average, however, the Bank has not responded optimally to each individual indicator. The Bank systematically overreacted to some indicators and underreacted to others. Correcting these errors could improve inflation targeting in Canada.