Using estimates of item ease and item discrimination, procedures are provided for computing estimates of the reliability and percentage of failing scores for tests assembled from these items. Two assumptions are made: that the average item coefficient will be approximately equal to the average of the estimated coefficients and that the score distribution for the test will be approximately normal. The predicted mean test score is equal to the sum of item ease coefficients, and the predicted variance is equal to the square of the sum of item discrimination indices divided by 4.5. The fail point is always sixty percent of the number of items. Normal curve tables are used to estimate the percent of score below the fail point. A normal curve probability table is provided, as is a computer program in BASIC for using this method. The results of using this procedure with seventeen tests are presented and compared with the obtained score distribution statistics for samples of either 51 cases or 201 cases. Kuder-Richardson formula 21 reliability coefficients may also be obtained from this procedure. (CTM)