Abstract: Objective: To investigate the feasibility of BP neural network model for predicting the dose of clinical red blood cells.Method: The monthly dose of clinical leukocyte-suspended red blood cells in Shunde District of Foshan City from January 2012 to December 2018 was collected.BP neural network model was used to predict the amount of clinical leukocyte-suspended red blood cells in our district from January 20 to September 2019 based on this data.Result: The absolute value of deviation between the predicted and actual values in February 2019 was relatively large,which was 27.6%,and the absolute value of deviation in the remaining months was 10%.Conclusion: The BP neural network can effectively provide a scientific basis for the blood collection and supply department to develop mining and blood supply plans,so as to better promote the rational use of blood resources and ensure the timely and efficient use of clinical blood.