To address the problem of weak echo signals and noise-induced instability in long-range millimeter-wave radar ranging, a novel ranging method combining an axially extended dielectric lens and a continuous peak tracking constant false alarm rate (CFAR-PT) algorithm is proposed. First, an axial extended planoconvex dielectric lens is designed, and its extension length is optimized using CST electromagnetic simulations to achieve beam compression and antenna gain enhancement. The lens is fabricated from polytetrafluoroethylene, with a diameter of 40.62 mm, a focal length of 30 mm, and an optimal extension length of 9 mm, operating in the 59~64 GHz frequency band. Second, to address the non-uniform distribution of echo energy along the range dimension caused by the dielectric lens, a CFAR-PT algorithm based on continuous peak tracking is proposed. Built upon conventional cell-averaging CFAR detection, the algorithm introduces the target index from the previous frame as a positional constraint. By performing neighborhood selection and amplitude comparison, the local maximum peak is identified, enabling stable target indexing. Finally, an experimental platform is built based on a millimeter-wave radar chip, and the system′s ranging accuracy is less than 2 mm. Mid-to-long-range comparison experiments are conducted in the 59~64 GHz frequency band. Simulation results show that the antenna gain exceeds 20.79 dBi across the entire frequency band, reaching 21.1 dBi at 61.5 GHz, which is 1.2 dBi higher than that of a conventional planoconvex lens. The 3 dB beamwidth is reduced from 8° to 6.7°. Experimental results demonstrate that both conventional CFAR and CFAR-PT achieve accurate detection in the mid-range region. However, in the long-range region, conventional CFAR fails due to false targets caused by strong near-range noise, whereas the proposed CFAR-PT algorithm effectively suppresses such interference through inter-frame positional constraints, enabling stable and reliable target tracking. The results demonstrate that the proposed co-design of lens structure optimization and detection algorithm significantly improves the accuracy and stability of long-range millimeter-wave radar ranging.