The invention relates to the technical field of lithium battery parameter determination methods, in particular to a lithium battery capacity rapid estimation method. The lithium battery capacity rapid estimation method comprises the following steps: (1) taking the-number-set training sample batteries to have an internal resistance test by using a composite pulse current method in order to obtain internal resistance values of the training sample batteries, (2) testing capacity values of the training sample batteries by using a constant current constant voltage test method, (3) taking time-set various current internal resistance of different sampling time of the training sample batteries as input and taking the capacity values as output to establish an internal resistance-capacity neural network model, (4) and testing internal resistance of predicting sample batteries and evaluating the capacity values by using the internal resistance-capacity neural network model established by the training sample batteries. By using the non-linear relationship between the internal resistance and the capacity to establish the internal resistance-capacity neural network model, the lithium battery capacity rapid estimation method can rapidly estimate the capacity of the batteries, eliminate the batteries which have no use value, reduce power loss and equipment loss required in a traditional capacity test, and greatly improve economy.