Abstract To assess the diagnostic applicability of magnetic resonance imaging (MRI) for diagnosis of cardiac allograft rejection, 25 patients who recently underwent cardiac transplantation were studied on a 0.15-tesla resistive system within 24 hours of endomyocardial biopsy. Ten normal volunteers and 4 patients who had recent (within 2 weeks) nontransplant cardiac surgery were also studied. In the 19 transplant patients imaged within 24 days of graft implantation, only 1 had evidence of graft rejection on biopsy. However, all nonrejecting grafts had increased T1 and T2 values, 501 ± 22 and 61 ± 6 ms, respectively (mean ± standard deviation) and the only rejecting graft had values of 496 and 60 ms, respectively. In the normal volunteers mean T1 was 352 ± 18 ms and T2 was 35 ± 6 ms. There was no significant difference in T1 and T2 values between patients who underwent nontransplant surgery and control subjects. In patients with nonrejecting transplants who were were imaged more than 25 days after surgery, the T1 and T2 values had normalized to 359 ± 17 ms and 36 ± 7 ms, respectively (n = 28 images in 20 patients). However, in those grafts with rejection, T1 and T2 were both elevated to 502 ± 21 ms and 62 ± 6 ms, respectively (n = 15 in 13 patients); wall thickness was also increased. Fourteen of 15 late rejection events (more than 25 days after surgery) were correctly identified on the basis of increases in T1 and T2 to more than 2 standard deviations above normal. Only 1 of 28 images of nonrejecting grafts was incorrectly identified as indicating rejection on the basis of prolonged T1 and T2 values. Thus, MRI appears to have potential as a noninvasive technique for diagnosis of cardiac allograft rejection when subjects are studied more than 25 days after surgery. Before that, all grafts showed increased T1 and T2 alues and wall thickness, presumably due to edema in response to graft harvesting, transportation and implantation.