To evaluate the initial results of predicting lymph node metastasis in rectal cancer patients detected invivo with USPIO-enhanced MRI at 3 T compared on a node-to-node basis with histopathology. Methods: Ten rectal cancer patients of all clinical stages were prospectively included for an in-vivo 0.85 mm 3 isotropic 3D MRI after infusion of Ferumoxtran-10. The surgical specimens were examined ex-vivo with an 0.29 mm 3 isotropic MRI examination. Two radiologists evaluated in-vivo MR images with a classification scheme to predict lymph node status. Ex-vivo MRI was used for MR-guided pathology and served as a key link between invivo MRI and final histopathology for the node-to-node analysis. Results: 138 lymph nodes were detected by reader 1 and 255 by reader 2 (p = 0.005) on in-vivo MRI with a median size of 2.6 and 2.4 mm, respectively. Lymph nodes were classified with substantial inter-reader agreement (κ = 0.73). Node-to-node comparison was possible for 55 lymph nodes (median size 3.2 mm; range 1.2-12.3), of which 6 were metastatic on pathology. Low true-positive rates (3/26, 11 % for both readers) and high true negative rates were achieved (14/17, 82 %; 19/22, 86 %). Pathological re-evaluations of 20 lymph nodes with high signal intensity on USPIO-enhanced MRI without lymph node metastases (false positives) did not reveal tumor metastasis but showed benign lymph node tissue with reactive follicles. Conclusions: High resolution MRI visualizes a large number of mesorectal lymph nodes. USPIO-enhanced MRI was not accurate for characterizing small benign versus small tumoral lymph nodes in rectal cancer patients. Suspicious nodes on in-vivo MRI occur as inflammatory as well as metastatic nodes.