This study uses iterative solution techniques to determine empirically optimal standards for student to faculty ratios by discipline and student level for a group of 11 public, research universities, all of whom were members of the Association of American Universities. One set of benchmark standards was based on four-variable solutions within disciplines where the four levels were lower division, upper division, master's and professional, and doctoral. A second set of benchmark standards was based on six-variable solutions that separately identified individual instruction and dissertation hours. These benchmarks were derived through a statistical process that found the set of standards that best described the relationship between student credit hours and instructional full-time equivalency personnel (IFTE) within discipline for the 11 institutions. Benchmark instructional personnel costs were computed by linking predicted IFTE and the known distribution of IFTE among these institutions with benchmark salary information. Both of the benchmarks were more accurate than state formula funding standards and both models were reasonably accurate. Possible applications of the benchmarks in considerations regarding hiring and resource allocation are noted. Two tables which present the data are appended. (SW)