Abstract This paper describes the results of a series of graduate-engineering classroom exercises in applying the Delphi procedures for formulating group judgments. The exercises were conducted in two phases. The first phase of the exercises was essentially a demonstration of the basic principles of the Delphi method using almanac, factual-type questions for which the answers are known. The second phase was a group exercise in formulating opinions about the educational goals of the United States for the 1970–1980 decade, using questions for which the true answers are not known and, therefore, value-judgments must be made. In the demonstration phase of the exercises, successive iterations with information feedback resulted in higher group-accuracies, when estimating the true numerical values of the answers to given questions. Furthermore, the individual numerical estimates exhibited a lognormal-distribution behavior. These findings are consistent with those obtained by other experimenters on applications of Delphi techniques. In the latter phase of the exercises, a consensus was readily obtained in regard to identifying the educational goals deemed most desirable and beneficial to the general welfare of the nation. However, considerable difficulty was experienced when value-judgment attempts were made by the group to categorize these goals according to four criteria: desirability, benefit, feasibility and cost. Here, also, our findings were consistent with those of other experimenters.