Since we do not live in an idealized world with infinite computational and monetary resources, the question of dollars worth of difference between ENDF/B cross-section processing codes for fast reactor calculations must be addressed with respect to both the accuracy required in the multigroup constants and the cost of obtaining that accuracy. Unfortunately, no single processing code is an obvious choice since the practical effect of different processing techniques on important reactor parameters is problem dependent* Therefore, it is the purpose of this paper to discuss the range of applicability of several multigroup averaging programs widely used in the United States, as well as the effort that must be expended in using them. Th° programs selected from the authors' experience include AMPX (1), MOIX( i 2 J .3)/SPHINX(4), ETOE-2/MC 2 -2/SDX(5), and ETGX(6) /1DX(2) • As is apparent these programs are often systems of linked computer codes and such distinctions will be highlighted in the discussion. Several articles (8-lfJ) have already been published which discuss the different techniques and computer programs employed in cross-section processing. Much of this work, however, is of a survey nature and without numerical detail sufficient to quantitatively contrast the differences between processing codes. Others (11)(12)(13)(14) are more quantitative but either do not involve all the programs considered herein, review only a limited number of reaction types, or contain integral results, but limited information on specific group-wise cress sections.In the next section, a brief summary is given of the status and current capabilities for the programs under discussion. This is followed by a section in which, some of the more important processed multigroup constants, obtained with the aforementioned codes (for a particular model of ZPR-6/7) are tabulated to illustrate the magnitude of the differences obtained for Individual cross sections and transfer matrix elements. It is essential to note that all of the programs involved have several options and no attempt was made to exploit these options to "fine-tune" or "optimize"