Patients with inflammatory heart muscle diseases would benefit from a safe, convenient, rapidly performed diagnostic technique with real-time results not involving tissue removal. We have performed a detailed evaluation of detection of heart allograft rejection by autofluorescence in a heterotopic abdominal rat heart allograft model ex vivo. Recipient rats with allograft (Lewis to Fisher 344; n=71) and isograft (Lewis to Lewis; n=33) hearts, treated with cyclosporine or untreated, were killed at days 2, 4, 7, 14, 21, 28, and 56 after transplant. Nontransplant controls with (n=24) or without (n=24) immunosuppressive therapy were also studied. When the rats were killed, autofluorescence spectra were acquired under blue-light excitation from midtransverse ventricular sections of native and transplanted hearts. Corresponding sections were then evaluated pathologically by a modified International Society for Heart and Lung Transplantation (ISHLT) grading schema. The spectral differences between rejecting and nonrejecting hearts were quantified by linear discriminant functions, producing scores that decreased progressively with increasing severity of tissue rejection. Mean+/-SD discriminant function scores were 2.9+/-1.6, 1.8+/-2.2, -0.1+/-2.8, -1.2+/-2.3, and -2.3+/-3.0 for isografts and allograft ISHLT grades 0, I, II, and III, respectively (Spearman rank-order correlation -0.6; P<0.001, test for trend). Cyclosporine had no detectable effect on the spectra. The correlation between changes in autofluorescence spectra and ISHLT rejection grade strongly supports the possibility of catheter-based, fluorescence-guided surveillance of rejection.