The effect of alternative census dates on the determination of student enrollment and retention rates and the accuracy of tuition revenue projections was analyzed at Indiana University-Purdue University Indianapolis. A set of pseudo-extract files was created for the first 8 weeks of two consecutive fall semesters (1993 and 1994). The accuracy of the files was determined by comparing student enrollment and credit hour counts for the end of the first week to the official student census file. Student enrollment counts and revenue projections were derived from each weekly file for the Fall 1993 semester as well as for the end of the semester. The retention rates were calculated by extracting a cohort from each weekly fall 1993 file and determining students' enrollment status from the corresponding fall 1994 weekly pseudo-census extract. Data extracted at the end of the third week of the semester generated the smallest combined deviations of the examined measures from their semester optimum values. Several other factors are identified that could influence the decision as to when official institutional data should be generated. The study demonstrates the usefulness of taking multiple criteria into account when determining the point-in-time at which to extract data for these vital management functions. Contains eight references and nine tables. (SW)