The development of new surgical techniques and use of neoadjuvant therapy have increased the need for accurate preoperative staging of rectal cancer. We compared the ability of endoscopic ultrasonography (EUS) and two magnetic resonance imaging (MRI) coils to locally stage rectal carcinoma before surgery. Forty-nine patients with histologically proven rectal carcinoma were T and N staged by EUS and either body coil MRI or phased-array coil MRI. After radical surgery, the preoperative findings were compared with histologic findings on the surgical specimen. For T stage, accuracies were 70% for EUS, 43% for body coil MRI, and 71% for phased-array coil MRI. For N stage, accuracies were 63% for EUS, 64% for body coil MRI, and 76% for phased-array coil MRI. For T stage, EUS had the best sensitivity (80%) and the same specificity (67%) as phased-array coil MRI. For N stage, phased-array coil MRI had the best sensitivity (63%) and the same specificity (80%) as the other methods. EUS and phased-array coil MRI provided similar results for assessing T stage. No method provided satisfactory assessments of local N stage, although phased-array coil MRI was marginally better in assessing this important parameter. Although none of the results differed significantly, phased-array coil MRI seems to be the best single method for the preoperative staging of rectal cancer.