In order to assess the recent trend of nonoperative management of pancreatic necrosis, we reviewed 82 variables in 73 consecutive patients with symptomatic necrotizing pancreatitis. The mortality rate for the series was 25% (18 of 73). The only preintervention variables that correlated with mortality were APACHE II score greater than 15 (p = 0.01), preintervention blood transfusion (p less than 0.001), respiratory failure (p less than 0.001), and shock (p less than 0.01). Patients who developed recurrent sepsis following the initial intervention had a significantly higher mortality rate (17 of 34) than those who did not (1 of 39) (p less than 0.001). The rate of recurrent sepsis varied widely among individual surgeons and correlated with APACHE II score. The presence of infected versus noninfected necrosis did not correlate significantly with outcome. When percutaneous radiologically guided drainage was the initial therapeutic modality (n = 6), recurrent sepsis requiring surgical drainage inevitably occurred. Patients treated with percutaneous drainage (often in combination with surgical drainage) had a longer hospital stay (82 versus 42 days, p less than 0.001), spent more days in the intensive care unit (31 versus 6 days, p less than 0.001), and required more days of total parenteral nutrition (57 versus 27 days, p less than 0.001) than those treated solely by surgical means. We conclude that aggressive initial surgical débridement should be the first step in managing symptomatic pancreatic necrosis and that the presence of infection should not be the sole determinant of intervention.