Abstract: Dynamic framed slotted Aloha algorithm is one kind of commonly used passive radio frequency identification (RFID) tag anti-collision algorithms. In the algorithm, the frame length requires dynamical set to ensure high identification efficiency. Generally, the settings of the frame length are associated with the number of tags and the probability of capture effect. Traditional estimation algorithms can estimate the number of tags and the probability of capture effect, but the number of tags is greater than an initial frame length when it is in dense RFID tags environment, and the estimation errors will increase significantly. In order to solve the problem that the conventional algorithms can not be applied to dense RFID tags environment, capture-aware Bayesian tag estimation is proposed in the paper, and the settings of optimal frame length with non-isometric slots are given. From the experimental results, the proposed algorithms have significantly lower estimation errors than traditional algorithms in dense RFID tags environment. And the identification efficiency got by setting the frame length according to the estimation results is also higher than that of traditional algorithms.