A prospective study comparing colour Doppler ultrasound (US) with the 'gold standard' of intra-arterial digital subtraction angiography in the evaluation of renal transplant artery stenosis was performed. Both the intrarenal vessels and the transplant renal artery were examined by Doppler US. Diagnostic arteriography was performed only if, on Doppler, the peak systolic velocity in the transplant renal artery exceeded 1.5 ms-1. Of 109 patients, the transplant artery could not be visualized using colour Doppler US in three, and these were omitted from statistical analysis. Of the remaining 106 patients, 31 had a peak systolic velocity greater than 1.5 ms-1 in the transplant renal artery and were referred for DSA. Of the multiple renal Doppler indices recorded, the peak systolic velocity in the transplant artery was the best discriminating measurement for the detection of renal artery stenosis. A peak systolic velocity of > or = 2.5 ms-1 in the transplant renal artery had a sensitivity of 100% and a specificity of 95% for the detection of renal artery stenosis ( > 50% diameter reduction). Although a significant difference in Pulsatility Index, Resistive Index, Acceleration Index and Acceleration Time was recorded from the intrarenal vessels in the angiographically normal and stenosed groups with Doppler, these measurements were less useful as discriminating diagnostic tests. In conclusion, the peak systolic velocity in the transplant renal artery is the most sensitive Doppler criterion for renal artery stenosis and is sensitive and specific enough to be used as a screening test. The intrarenal acceleration time and index should not be used in isolation.