Abstract: In this paper a novel particle filter based track-before-detect algorithm, namely a particle filter based incoherent integration method for detection, is proposed for detection of weak target. In the proposed method, a target was declared to be present if the maximum accumulation of the absolute measurements along the possible tracks exceeds a given threshold, where the possible tracks were searched according to the resampling procedure in particle filter. The estimated state sequence of target could be output as the target is declared to be present. The detection performance of the proposed method was well enough and could be analyzed via extreme value theory. Thus the explicit expressions such as the probabilities of detection and false alarm and detection threshold can be obtained. An example of sky wave over-the-horizon radar target detection and tracking is present to demonstrate the capability and efficiency of the proposed method.