Abstract: Millimeter-wave cloud radar serves as a high-resolution, all-weather, ground-based remote sensing instrument used for cloud vertical structure detection. It can effectively compensate for spatiotemporal coverage limitations inherent in other cloud measurement methods. However, its detection accuracy can be affected by various complex factors, including climatic conditions, geographical environment, and instrument calibration. Meanwhile, L-band radiosonde soundings can provide accurate depictions of atmospheric temperature and humidity profiles, enabling the derivation of cloud vertical structure information, which serves as a valuable reference for assessing the accuracy of millimeter-wave cloud radar detection. Cloud height data derived from L-band radiosonde soundings at 59 co-located sites with millimeter-wave cloud radar across China from January to December in 2024, are used as a benchmark in this study. By establishing matching rules based on temporal and spatial proximity between two datasets, the study systematically assesses the consistency of cloud height (including cloud base and cloud top height) and cloud layer number between millimeter-wave cloud radar and L-band radiosonde soundings. Results indicate that millimeter-wave cloud radar and radiosonde data are highly consistent in identifying the presence or absence of clouds, demonstrating a consistency rate of 78.95%. In detecting cloud base height, the millimeter-wave cloud radar demonstrates high accuracy, with a mean error of -0.21 km and a correlation coefficient of 0.89 (passing the test of 0.001 level). For cloud top height, the millimeter-wave cloud radar tends to underestimate compared to L-band radiosonde soundings, exhibiting a mean error of -0.92 km and a correlation coefficient of 0.84 (passing the test of 0.001 level). It’s also found that millimeter-wave cloud radar shows higher accuracy in detecting mid-level clouds compared to low-level and high-level clouds. Temporal analysis demonstrates that assessment indicators remain relatively stable across different months, with the largest errors occurring from July to September. During this period, the correlation coefficient for cloud base height falls below 0.85, with a root mean square error of approximately 1.5 km. The correlation coefficient for cloud top height is lower than 0.8, with a root mean square error of about 2 km. Spatially, regional differences are observed: Mean errors in cloud base height are predominantly negative in the northern regions and positive in the southern regions, while mean errors in cloud top height are generally negative, with smaller magnitudes in inland areas compared to the northern and coastal regions. These spatiotemporal distribution patterns are possibly attributable to differences in climatic conditions (e.g., precipitation patterns and humidity distribution) between the north and the south, as well as geographical environments (e.g., topographic relief, maritime influence, and aerosol concentration distribution). Overall, millimeter-wave cloud radar performs well in detecting cloud vertical structure and shows high consistency with L-band radiosonde soundings. Results are practical, though there are some identified systematic biases.