In a world with uncertainty where foresight is imperfect, the well-known Fisher relation can be interpreted as an equality between nominal rate of interest and the equilibrium expected real return plus the market's assessment of the expected rate of inflation. This relationship has been subjected to extensive econometric tests, including those by Fisher [6] himself, utilizing distributed lags of past inflation rates as the observable proxy for the price expectations variable. The efficient market literature, on the other hand, utilizes the rational expectations hypothesis (REH) which states that market's anticipated value equals the model's expected value of the variable being predicted. In a recent article, Fama [4] has produced overwhelming evidence in favor of market efficiency with some reservations in respect to five and six month treasury bills.' However, since Fama's tests are joint tests of the REH, full Fisher effect, and the constancy of real rates, one cannot readily identify the particular hypotheses being invalidated. The main purpose of the present paper is to test the rational expectations hypothesis and the Fisher effect as two disjoint propositions. In addition, to highlight the implications of different expectational proxies on the estimated Fisher effect, we have used the well-known Livingston's series on price expectations which has recently been adjusted by Carlson [2]. Using the original Livingston's data, Cargill [1] has recently reported estimates of the Fisher effect in the treasury bill rates of different maturities. We, however, used the survey data in an errors-in-the variables framework and let the model decide the significance of the hypothesized sampling errors. Moreover, the use of Livingston's data in interest rate equations of different maturities typically produces the price expectations coefficients which are significantly more than one. Hendershott and van Home [11] have recently produced some evi-