PURPOSE: To evaluate the ability of magnetic resonance (MR) imaging to depict solid debris within pancreatic collections prior to intervention and to help assess drainability, as well as to compare MR findings with those obtained at computed tomography (CT) and ultrasound (US). MATERIALS AND METHODS: Nineteen collections in 18 patients were evaluated with MR imaging, CT, and US prior to drainage. Prospective, blinded interpretations of imaging studies by three independent readers (each interpreted all the images obtained with only one modality) evaluated collection characteristics (debris, consistency, septation, wall thickness, and irregularity) and predicted drainability. Findings were compared with clinical diagnosis and clinical outcome of drainage. RESULTS: MR imaging and CT depicted all collections; US failed to depict two collections. In nine patients with subacute necrotic collections, solid debris was seen in eight (89%) at MR imaging, in two (22%) at CT, and in eight (89%) at US. In seven patients with pseudocysts, debris was seen in two (28%) at MR imaging and in none at CT, as well as in six (100%) of six at US. A collection was defined as 'not drainable' on the basis of the depiction of solid necrotic debris more than 1 cm in diameter. With this definition, statistically significant differences between sensitivity and specificity values, respectively, were found for the prediction of actual drainability: MR imaging, 100% and 100%; CT, 25% and 100%; US, 88% and 54%. CONCLUSION: Predrainage MR imaging should be performed in patients with subacute pancreatic collections to avoid infectious complications from unrecognized necrotic debris that cannot be removed with use of standard pseudocyst drainage techniques.