An interactive FORTRAN program, MUDIFT, is presented. This program was designed to perform a multivariate distribution-free significance test for the comparison of growth curves. Such a test, not assuming functional forms of individual growth, has proved useful when the variety of observed growth curves was too broad to be represented by the family of growth functions and when there were incomplete observations. A numerical example illustrates the use of MUDIFT to analyze serial data pertaining to the measurable solid tumors in 20 mice treated with liposome incorporated muramyltripeptide phosphatidylethanolamine and a control group of 30 mice. The tumor volumes were recorded weekly for 10 weeks following initial entry into the research protocol. The test hypothesis was whether the tumors grew slower in the treated group than in the control group. The test statistic, an asympotically chi 2 statistic, was 7.312 with df = 2, which was significant at alpha = 0.05.