This paper analyses Hungarian inflation expectations drawn from Reuters survey of professional forecasters. Comparing individual forecasts to macroeconomic consensus leads to the conclusion that - in line with theoretical considerations - averaging single projections results in significant gains in terms of forecast accuracy. Econometric tests reveal that consensus forecast is an unbiased estimation of actual inflation, although autocorrelation in forecasts errors and revisions contradicts the requirement of efficiency. Taking into account the varying sample and the heterogeneity of particular forecasts, minor changes in consensus (0.1-0.2 percentage points) usually may not be considered as an unambiguous sign of altering inflation expectations. On the other hand, three consecutive changes in the same direction are almost surely the consequence of shift in actual expectations.