Objectives: To distinguish between primary mild COVID-19 patients and the patients with recovery positive virus detection after discharge (recovery positive patients), the clinical characteristics and biochemical indicators were analyzed. Methods: By evaluation of correlations between biochemical indicators and serum ferritin (FE) through univariate and multivariate analyses, we evaluated the differences of FE index and analyzed the effectiveness of the FE* pre-albumin (PA) on discrimination between primary mild and recovery positive COVID-19 patients. The area under curve (AUC) and a predictive nomogram were further employed to evaluate the effectiveness of discrimination. Results: In comparison with recovery positive patients, liver damage-related indicators such as serum FE and PA were significantly lower in primary COVID-19 patients with mild symptoms. Area under the ROC curve values of FE, PA and FE*PA were 0.81, 0.78 and 0.83, respectively. In combination with multivariate analysis and nomogram, FE and FE*PA can be considered as reasonable evaluation systems for predicting primary mild and recovery positive patients. Conclusions: FE*PA was an independent predictive biomarker for distinguishing primary mild and recovery positive COVID-19 cases.