A study was conducted at the University of Maryland to determine how consistently students perform on examinations in a given class and whether it is possible to alter this relationship through the introduction of an intervening variable, such as a reinforcing and encouraging letter from the instructor. Letter grades and test scores were obtained for all 315 students in an introductory psychology class in the Fall of l969. Prior to the second exam, the 7l students who received grades of A or B on the first exam were randomly assigned to two groups, and 35 students in one of the groups received a letter from the instructor commending their performance and indicating expectations of similar subsequent performance. Correlations among the first three exams were computed for students receiving A's and B's on the first exam and the group receiving C's, D's, or F's, and for those in the A and B group who did and did not receive letters. Scores on subsequent tests were more highly related to the first exam for the group of students who received A's or B's on the first test than the group which received C's, D's, or F's. The difference was not great, and interpretations must consider the limitation in range of the A and B group, which may result in lower correlations than the more heterogenous condition of the C, D, and F group. The distribution of "Reinforcement" letters to a sample of students who received A's or B's on the first exam had no observable effect on subsequent test performance. (SW)