Traditional deinterleaving algorithms first cluster the pulses in frequency, pulse width, and angle of arrival parameter space. Then, they utilize time of arrival analysis techniques. This results in a performance loss, since these processes are not jointly combined; rather, they are sequentially connected. Additionally, they perform poorly in the presence of radars with staggered pulse repetition interval type due to abrupt time difference variations. To alleviate the suboptimality in the design and enhance the detection performance in the staggered type, we propose a combined approach based on the pulse repetition interval transform technique. It does not cluster the pulses as the traditional methods, instead, it combines all pulses by distance weighting. It deals with the detection problem in the staggered type by clustering the phases of the autocorrelation function. We develop an optimal weighting mechanism to fuse the outputs from phase clusters. Numerical results show that our method outperforms the traditional methods in different detection metrics via multiple scenarios, which is also validated by the theoretical findings.