When there is uncertainty about estimates of the margin of unused capacity in the economy, examining a range of inflation indicators may help in assessing the balance of risks regarding the outlook for inflation. This paper tests a wide range of observable variables for their leading-indicator properties with respect to core inflation, including: commodity prices, cost indicators, measures of capacity pressures in labour and product markets, and components of the consumer price index (CPI) itself. After a preliminary screening of indicators using Granger causality tests, estimated bivariate indicator models generate post-sample static forecasts one quarter ahead and two quarters ahead over the period 1995(Q1)–1999(Q1). A ridge regression technique is used to optimally combine selected bivariate forecasts into multivariate forecasts. The root-mean-squared errors of both the bivariate and multivariate forecasts are compared with those of benchmark models—a Phillips curve, an autoregressive model, and two naive models. The results show that several indicator models generate lower forecast errors than the benchmark models over the post-sample period. Several CPI components, as well as the Bank of Canada commodity price index in U.S. dollars, the industrial product price index for electrical products, the average prices for resale housing in four major cities, and the ratio of unfilled orders to shipments in manufacturing are among the best predictors of core inflation. The paper also briefly discusses the limitations of indicator models, including the possibility that predictions from such models may not tell us much about the underlying pressure of demand on production capacity or the fundamental trend in inflation.