Purpose: To determine whether treatment with Plasmalyte-148 (PL) compared to sodium chloride 0.9% (SC) results in faster resolution of diabetic ketoacidosis (DKA) and whether the acetate in PL potentiates ketosis. Methods: We conducted a cluster, crossover, open-label, randomized, controlled Phase 2 trial at seven hospitals in adults admitted to intensive care unit (ICU) with severe DKA with hospital randomised to PL or SC as fluid therapy. The primary outcome, DKA resolution, was defined as a change in base excess to ≥ − 3 mEq/L at 48 h. Results: Ninety-three patients were enrolled with 90 patients included in the modified-intention-to-treat population (PL n = 48, SC n = 42). At 48 h, mean fluid administration was 6798 ± 4850 ml vs 6574 ± 3123 ml, median anion gap 6 mEq/L (IQR 5–7) vs 7 mEq/L (IQR 5–7) and median blood ketones 0.3 mmol/L (IQR 0.1–0.5) vs 0.3 (IQR 0.1–0.5) in the PL and SC groups. DKA resolution at 48 h occurred in 96% (PL) and 86% (SC) of patients; odds ratio 3.93 (95% CI 0.73–21.16, p = 0.111). At 24 h, DKA resolution occurred in 69% (PL) and 36% (SC) of patients; odds ratio 4.24 (95% CI 1.68–10.72, p = 0.002). The median ICU and hospital lengths of stay were 49 h (IQR 23–72) vs 55 h (IQR 41–80) and 81 h (IQR 58–137) vs 98 h (IQR 65–195) in the PL and SC groups. Conclusion: Plasmalyte-148, compared to sodium chloride 0.9%, may lead to faster resolution of metabolic acidosis in patients with DKA without an increase in ketosis. These findings need confirmation in a large, Phase 3 trial.