Tissue diagnosis is important during surgical excision of solid tumors for margin evaluation. Conventional histopathologic methods rely heavily on image-based visual diagnosis by specialized pathologists, which can be time-consuming and subjective. We report a three-dimensional (3D) histological electrophoresis system for rapid labeling and separation of the proteins within tissue sections, providing a more precise assessment of tumor-positive margin in surgically resected tissues. The 3D histological electrophoresis system uses a tumor-seeking dye labeling strategy to visualize the distribution of tumor-specific proteins within sections and a tumor finder that automatically predicts the tumor contour. We successfully demonstrated the system's capability to predict the tumor contours from five murine xenograft models and distinguish the tumor-invaded region of sentinel lymph nodes. Specifically, we used the system to accurately assess tumor-positive margins from 14 patients with cancer. Our 3D histological electrophoresis system serves as an intraoperative tissue assessment technology for more accurate and automatic pathologic diagnosis.