The feasibility of applying magnetic resonance imaging (MRI) for conducting prospective studies of intraperitoneal (i.p.) tumor treatment response to chemotherapy and resultant effects on survival in human ovarian carcinoma/nude mouse orthotopic xenograft models was evaluated. Female nude mice were implanted i.p. with either NMP-1 or SKOV-3ip. human ovarian carcinoma cells on day 0. Initial T2-weighted magnetic resonance (MR) images of the abdomens of NMP-1-implanted mice were obtained on day 7 to confirm the presence of nascent tumors; similar confirmations were made on day 14 with mice bearing SKOV-3ip. xenografts. On the initial imaging days, a multiple-dose regimen of cisplatin (CDDP; qd7 x3) was commenced, using 4 or 6 mg/kg treatments with the NMP-1 model and using 6 mg/kg treatments with the SKOV-3ip. model. Mice were reimaged multiple times, 2 days following each CDDP injection and at later times as well, depending on host survival. The images for each mouse from the last imaging day (day 30 for NMP-1, day 44 for SKOV-3ip.) were used in a blinded fashion to attempt to visually distinguish control from treated mice and to determine whether MRI could predict a survival benefit. For SKOV-3ip. mice, ten out of ten mice were correctly segregated into the control or the CDDP treatment group based solely on these blinded, nonquantified MR results. In this model, the 6 mg/kg multiple-dose regimen achieved a modest response, improving life span by approximately 24%. However, for the NMP-1 mice, only six out of nine evaluable mice were correctly segregated into the control or one of the treatment groups by similar MRI criteria, a virtually random distribution; further, neither CDDP treatment regimen achieved a significant improvement in survival in this model. In another study, NMP-1-implanted mice were treated on day 7 after tumor implantation with a single injection of a hyaluronic acid-paclitaxel copolymer. Control and treated mice were MR imaged on day 28, which revealed marked reductions in tumor burden in treated mice, correlating well with a subsequently observed improved survival of approximately 40%. Our results suggest that MRI can be used to serially and noninvasively monitor treatment response and predict ongoing treatment effects on survival.