Background: Increasingly, peripancreatic fluid collections (PFCs) are managed endoscopically with conventional transmural drainage (CTD). The role of interventional EUS in drainage of PFCs requires further clarification, because the procedure is technically challenging, with limited availability. Objective: Identify characteristics that determine the need for drainage of PFC by CTD versus EUS. Patients: Consecutive patients with symptomatic PFCs (types: pseudocyst, abscess, and necrosis) referred for endoscopic drainage. Study Design: Prospective study. Setting: Tertiary-referral center. Methods: After ERCP, transmural drainage was attempted by CTD. If unsuccessful, drainage by EUS was then attempted. Findings on contrast-enhanced CT and endoscopy were collected to identify characteristics that predict the need for CTD versus EUS drainage. Main Outcome Measurements: Identify characteristics to determine whether CTD or EUS is best suited for drainage of a particular PFC. Technical outcomes and safety of both techniques were also compared. Results: Of 53 patients with PFCs, CTD was technically successful in 30 (57%) and failed in 23 (43%). PFC regional location was the pancreatic head in 16, the body in 20, and the tail in 17; in these locations, CTD was successful in 13 (81%), 17 (85%), and 0, respectively. The causes of failed CTD were absence of luminal compression (LC) in 20, difficulty with scope positioning in 2, and bleeding with attempted drainage (portal hypertension) in 1. One PFC drained by CTD was later diagnosed as necrotic sarcoma. Of the 23 patients who failed CTD and underwent EUS, an alternate diagnosis of mucinous neoplasm was made in 2 patients, and EUS-guided drainage was successful in the other 21 patients (100%). Although CTD failed in all PFCs in the tail, all were successfully drained by EUS. In the pancreatic-head region, only those PFCs superior to pancreas and extending into porta hepatis (n = 3) required drainage by EUS. In the pancreatic body, only PFCs that developed bleeding from a transmural puncture or without definitive LC because of gastric mural edema (albumin <1.5 mg/dL, n = 2) required EUS drainage. When compared with PFCs at other locations, those in the tail were best accessed by EUS (P < .001). Patients with luminal compression at CT were significantly more likely to undergo successful drainage by CTD (adjusted odds ratio [OR] 13.6; P = .02). When compared with CTD, EUS drainages were longer in duration (40 versus 75 minutes; P < .001), with similar rates of PFCs resolution (90% versus 95%). Although bleeding occurred in 1 patient in the CTD group, no complications were encountered in patients who underwent EUS-guided drainage. PFCs located at the tail of the pancreas were more likely to require drainage by EUS than CTD (adjusted OR 22.9, P = .003) when adjusted for the presence of luminal compression at CT, size of the PFC, serum albumin, and etiology of pancreatitis. Limitations: Nonrandomized study. Conclusions: Because a majority of PFCs can be drained by CTD in a shorter duration, with comparable outcomes, EUS-guided drainage should be reserved mainly for PFCs located at the pancreatic tail, because these are unlikely to cause luminal compression or are technically difficult to access. Also, all pseudocyst-type PFCs must be evaluated by EUS before any attempts at endoscopic drainage, because EUS identifies an alternate diagnosis in 5% of such patients. © 2007 American Society for Gastrointestinal Endoscopy.