To compare the results of the preoperative workup consisting of endoscopic retrograde cholangiography (ERC), magnetic resonance cholangiography (MRC), and percutaneous resonance cholangiography (PTC) with the tumor extent of the surgical specimen in patients with hilar cholangiocarcinoma (hilCC).Between 9/1997 and 12/2002, 59 patients with hilCC tumor underwent surgical resection. Preoperative ERC, MRC, and PTC were analyzed, blinded for the identity of the patient, and compared with the surgical specimen. For this retrospective analysis, 55 of the initial 59 ERCs, 39 of the initial 40 MRCs and 32 of the initial 38 PTCs were available. Most of the ERCs and MRCs had been performed at referring institutions by various investigators. In 20 patients, all three imaging modalities were available for direct comparison.The mean scores of the visualization of the bile ducts and tumor differ considerably for ERC, MRC and PTC, with PTC visualizing the bile ducts better than ERC (p < 0.001) and MRC (p = 0.019). The tumor classification according to Bismuth and Corlette was correctly predicted by ERC in 29 %, by MRC in 36 % and by PTC in 53 %. The tumor extent was overestimated in 40 % (ERC), 41 % (MRC) and 31 % (PTC) and underestimated in less than 10 % for all modalities. Twenty patients, who underwent all three imaging modalities, were included in an additional analysis for a direct comparison of ERC, MRC and PTC. PTC provided correct or acceptable information on tumor extent in 19 of 20 patients, MRC in 15 of 20 patients, and ERC in only 11 of 20 patients. The statistical analysis revealed a significant superiority of PTC to ERC (McNemar test: p < 0.01) but not to MRC (p = 0.22).The management of patients with hilar cholangiocarcinoma requires a high degree of expertise in diagnostic imaging techniques. Cholangiography should not only define the location but also visualize the uppermost extent of the tumor to determine resectability. In contrast to most reports in the literature, ERC and MRC were found to be of limited reliability regarding the assessment of the tumor extent. ERC may be more and more reserved for patients considered for nonsurgical intervention or palliation. PTC proved to be the most reliable approach. MRC represents a noninvasive diagnostic tool for the evaluation of malignant perihilar biliary obstructions, but should be performed at highest quality using state-of-the-art MRI techniques. The most common mistake of each diagnostic modality was an overestimated tumor extent, which may exclude patients from potentially curative surgery.