It is claimed in this book that expectations should not necessarily be treated as unobservable variables and that there is much to be learned from survey data. A unique data set is examined, the output of surveys conducted twice a year since 1952, among informed Italian businessmen and economic experts. The predictive accuracy, rationality and determinants of inflation expectations are investigated, following an extensive analysis of measurement issues.The estimate of inflation expectations are evaluated for both wholesale and consumer price changes, comparing them with those held by respondents to other surveys for different countries and with the forecasts generated by alternative predictors of the inflation process. The expectations considered in the study are shown to be remarkably accurate, anticipating all major price changes, even if during the years of high and rising inflation which have followed the first oil crisis they appear to underestimate on a number of occasions the inflation rates actually experienced, as the alternative predictors also do.An accurate testing of the rational expectations hypothesis is conducted, rejecting it over the entire sample period but not for the period of mild, but variable inflation which preceded the first oil crises.It is shown that a mixed adaptive-regressive model, with both error-learning and return-to-normality components adapts very well to the data considered in this study and that inflation expectations are also influenced by an uncertainty component which affects the adaptive coefficient. Furthermore, regression towards normality is slowed down when industrial capacity is utilized above normal, and vice-versa. Many other issues such as the dispersion of individual answers, the problems of aggregation and measurement error are also considered and an extensive bibliography of other works where use is made of direct information on expectations, is included.