Abstract: As a typical electronic countermeasure, RF masking pulses generate a high-density distribution of interference in the electromagnetic space, which severely degrades the performance of radar emitter signal sorting. To address this issue, this paper proposes a two-stage clustering-based sorting method based on density adaptation and probabilistic modeling. First, the Hierarchical Density-Based Spatial Clustering of Applications with Noise (HDBSCAN) algorithm is employed to perform initial sorting on mixed data containing RF masking pulses. This step enables adaptive modeling of non-uniform density structures, effectively removes noise and masking pulse interference, and yields an initial clustering structure of genuine radar signals. Subsequently, a Gaussian Mixture Model (GMM) is introduced to refine the initial clustering results. The model parameters are optimized using the Expectation–Maximization (EM) algorithm, thereby improving the sorting accuracy. Furthermore, a spatial indexing structure is incorporated to accelerate the computation of probability densities, reducing the overall computational complexity of the algorithm. Experimental results demonstrate that the proposed method effectively suppresses globally dispersed random masking pulses and maintains high average sorting accuracy with limited performance fluctuations in high-overlap structured pseudo-cluster masking scenarios, while also preserving computational efficiency.