The disclosure discloses a statistical method and apparatus for webpage access data. The method includes that: an initial dataset of user-accessed page behaviors is collected; first-time access data of a user is extracted from the initial dataset; the first-time access data is judged by using a pre-set judgement model to obtain a judgement result; when the judgement result is contrary to a pre-set result of the pre-set judgement model, dating-back matching is performed in the initial dataset by using the first-time access data to obtain associated access data; and the initial dataset is spliced and repaired based on the associated access data to obtain a repaired dataset. By means of the disclosure, the problem in the traditional art that the user identification accuracy is dissatisfactory due to the loss of Cookie information is solved.