Objective: To compare the effect of balanced multielectrolyte solutions (BMESs) versus normal saline (NS) for intravenous fluid on chloride levels and clinical outcomes in patients with predicted severe acute pancreatitis (pSAP) Background: Isotonic crystalloids are recommended for initial fluid therapy in acute pancreatitis, but whether the use of BMES in preference to NS confers clinical benefits is unknown. Methods: In this multicenter, stepped-wedge, cluster-randomized trial, we enrolled patients with pSAP (acute physiology and chronic health evaluation II score >= 8 and C-reactive protein >150 mg/L) admitted within 72 hours of the advent of symptoms. The study sites were randomly assigned to staggered start dates for a one-way crossover from the NS phase (NS for intravenous fluid) to the BMES phase (sterofudin for intravenous fluid). The primary endpoint was the serum chloride concentration on trial day 3. Secondary endpoints included a composite of clinical and laboratory measures. Results: Overall, 259 patients were enrolled from 11 sites to receive NS (n = 147) or BMES (n = 112). On trial day 3, the mean chloride level was significantly lower in patients who received BMES [101.8 mmol/L (SD: 4.8) vs 105.8 mmol/L (SD: 5.9), difference -4.3 mmol/L (95% CI: -5.6 to -3.0 mmol/L) ;P < 0.001]. For secondary endpoints, patients who received BMES had less systemic inflammatory response syndrome (19/112, 17.0% vs 43/147, 29.3%, P = 0.024) and increased organ failure-free days [3.9 days (SD: 2.7) vs 3.5 days (SD: 2.7), P < 0.001] by trial day 7. They also spent more time alive and out of the intensive care unit [26.4 days (SD: 5.2) vs 25.0 days (SD: 6.4), P = 0.009] and hospital [19.8 days (SD: 6.1) vs 16.3 days (SD: 7.2), P < 0.001] by trial day 30. Conclusions: Among patients with pSAP, using BMES in preference to NS resulted in a significantly more physiological serum chloride level, which was associated with multiple clinical benefits (Trial registration number: ChiCTR2100044432).