The Livingston survey data have been subjected to numerous tests of the rationality of expectations. The empirical strategies of each of these pre vious efforts is, however, flawed in one way or another. Here we remedy these shortcomings by recognizing and correcting for the problems of overlap ping forecast intervals and heteroscedasticity in the forecast errors. The principal empirical findings are that the hypotheses of unbiasedness and efficiency can be rejected for the CPI forecasts but not for the PPI fore casts. There seems also to be some tendency for survey panel members to underestimate the inflationary effects of recent expansion in forecasting the CPI and PPI.