〔Abstract〕 Objective To investigate the factors influencing the purity of peripheral blood natural killer(NK) cellsculturedin vitroand to establish a prediction model for the purity of peripheral blood NK cells. Methods The peripheral blood of 93 healthy donors was collected, the purity of NK cells was detected by flow cytometryafterin vitroculture, and the clinical physical examination indicators of the donors were collected. 77 cases were randomly selected as the training set, and the remaining 16 cases were used as the test set. Pearson′s correlation test was used to analyze the indexes related to the purity of NK cells, and multivariate regression analysis was used to establish a model for predicting the purity of NK cells. We analyzed the correlation between model predicted purity and actual purity and the area under the receiver operating characteristic curve(ROC) of the model. Results The indicators correlated with the purity of NK cells were age, triiodothyronine(T3), red blood cell distribution width(RDW), red blood cell volume distribution width standard deviation(RDW-SD), creatinine(Cr), lymphocyte percentage(LY%), glucose(Glu), percentage of eosinophils(EOS%), number of eosinophils(EOS) and platelet volume distribution width(PDW). For regression model, NK index was-164.557+2.544 RDW-SD +3.730 PDW +4.389 Glu +10.237 T3(R2=0.494,P<0.05). In the test set, the coefficient of determination(R-Square) between the predicted value and the true value of the NK index model was 0.725,Pwas 0.001, and the area under the ROC curve of the NK index model in the training set and test set was 0.815 and 0.938. Conclusion The NK index model can better predict the purity of peripheral blood-derived NK cells culturedin vitro, and provide theoretical guidance for the subsequent formulation of clinical NK cells reinfusion treatment plans.