Abstract: Objective To evaluate the efficacy of harmless acute pancreatitis score (HAPS) on admission for predicting mild acute pancreatitis (MAP). Methods We prospectively collected the data from consecutive AP patients admitted to West China Hospital between January, 2016 and August, 2017, and HAPS scores were calculated on admission. The clinical outcomes of the patients with harmless AP (HAPS>2) and those with non-harmless AP (HAPS≤2) were compared. MAP was defined based on the severity classification of 2012 Revised Atlanta guidelines. The area under the curve (AUC) of the receiver- operator characteristic curve (ROC), specificity, sensitivity, positive predictive value (PPV) and negative predictive value (NPV) of HAPS on admission for predicting MAP were analyzed. Results Of the 703 patients with AP analyzed, 182 were predicted to have harmless AP and 521 to have non- harmless AP, and the patients in the latter group had significantly worse clinical outcomes (P<0.001). The total number of patients with MAP was 359 in the study. The specificity, the sensitivity, the PPV and NPV of HAPS on admission for predicting MAP was 97.7% (95% CI: 95.4-99.0), 48.2% (95% CI: 42.9-53.3), 95.6% (95% CI: 91.5- 98.1) and 64.1% (95% CI: 59.8- 68.2), respectively, and the AUC was 0.749 (95% CI: 0.72- 0.78). Conclusion HAPS score on admission can accurately predict MAP.