Abstract Refeeding after acute pancreatitis (AP) is traditionally started in a successively increasing manner when abdominal pain is absent and pancreatic enzymes are decreasing. We aimed to evaluate length of hospital stay (LOHS) and refeeding tolerance for early refeeding and/or immediately full caloric intake in patients recovering from AP. Methods In this randomized, open-label trial, patients with AP were randomized into four different refeeding protocols. Group 1 and 2 received a stepwise increasing diet during three days while 3 and 4 received an immediately full caloric, low fat diet. Group 2 and 4 started refeeding early (once bowel sounds returned) and 1 and 3 started at standard time (bowel sounds present, no abdominal pain, no fever, leucocytes and pancreatic enzymes decreasing). Main outcomes measurements were LOHS and tolerance (ability to ingest >50% of meals without severe pain, nausea or AP relapse). Results Eighty patients were evaluated and 72 randomized (median age 60 years, range 24–85, 33 male). LOHS was significantly reduced after early refeeding (median 5 versus 7 days ( p  = 0.001)) but not in patients receiving immediately full caloric diet, compared to standard management (6 versus 6 days ( p  = 0.12)). There was no difference in refeeding tolerance comparing immediately full caloric diet versus stepwise increasing diet (31/35 (89%) versus 33/37 (89%) patients tolerating the treatment, p  = 1.00) or early versus standard time for refeeding (33/37 (89%) versus 31/35 (89%), ( p  = 1.00)). Conclusions Refeeding after AP when bowel sounds are present with immediately full caloric diet is safe and well tolerated. Early refeeding shortens LOHS.