Hanldn, J. H. (School of Public Health, University of Hawaii, Honolulu, Hawaii 96822), H. B. Messinger and R. A. Stallones. A short dietary method for epidemi-ologic studies. IV. Evaluation of questionnaire.Amer. J, Epid., 1970, 97: 562–567.—The efficiency of a short recall questionnaire for predicting total dietary intakes was tested in a group of 55 Japanese-American men in 1967. Multiple regression equations, computed from a 1965 study of the same men, were used to predict individual daily intakes of total calories, protein, fat, carbohydrate, and sodium from the reported frequencies of 23 food groups. Comparison of the predicted values with measured intakes in 1967 revealed that differences of 10% or lets occurred in a range from 12 persons for carbohydrate to 21 for calories, and those of 20% or less in a range from 27 persons for sodium to 40 for calories. The 1965 equations showed a general tendency to overestimate the measured intakes In 1967 and a relative tendency to overestimate the low and underestimate the high intakes. Stepwise regression analysis of the 1967 measured intakes on the frequencies reported in the questionnaires resulted in lower multiple R's than the 1965 equations. Except for sodium, the values were fairly dose. Comparison of the 1965 and 1967 measured intakes revealed a general decrease for the group and also some regression towards the mean for persons with extremely high or low intakes. Procedures for improving the prediction equations and for using the short method in other populations are described.