The vertical distribution of aerosols has important implications on haze formation as development, which is manifested to some extent by the planetary boundary layer (PBL)–aerosol interactions. Information on the number concentration and size of particles is essential to understand these processes, but studies on vertical profiles of particle number-size distribution are limited. Herein, an unmanned aerial vehicle (UAV) equipped with a custom-built optical particle counter (0.4–10 μm) was used to investigate the vertical profiles of particle number-size distribution in Hefei (China) during January 20–30, 2021. Combining ground-based scanning mobility particle sizer and meteorological data, the pollution accumulation and diffusion mechanisms were analyzed in depth. Results showed that as the pollution episode developed, the vertical distribution of the particle number concentration changed from a flat profile to a sharp vertical gradient. Under polluted conditions, a three-layer structure was clearly evident: uniform distribution in a mixed layer near the ground, a sharply reduced transition layer, and a low number concentration layer in the free atmosphere. Analysis revealed that fundamental to this conversion is that aerosols are highly affected by the PBL dynamics. Concurrent on-UAV and ground-based observations revealed that the ratio of particle numbers in the accumulation mode to that in the Aitken mode was 0.92 ± 0.05 in polluted days, which was almost three times that of clean days. This difference in the ratio of large to small particles suggests that hygroscopic growth of aerosol particles under high humidity conditions played an important role in haze development. Moreover, the sharp vertical gradient of the particle number concentration in the transition layer was identified as an important parameter for characterizing PBL height. The findings in this study highlight the importance of PBL dynamics on the under-studied vertical profiles of particle number-size distribution, especially during heavy pollution episodes.