This study aims to investigate the spatiotemporal variation, chemical characteristics and source apportionment of marine fine particles (PM2.5), and to analyze the transport route towards the intersectional region of the Taiwan Strait and South China Sea. Sampling of PM2.5 was conducted simultaneously at Penghu and Dongsha Islands for 14 consecutive days each quarter from summer 2019 to spring 2020. PM2.5 samples were then returned for conditioning, weighting, and chemical analysis. The chemical mass balance (CMB) receptor model was used to identify the potential PM2.5 sources. It results show that the lowest PM2.5 concentrations were observed in summer, with a gradual increase starting fall, influenced by Asia Northeastern Monsoons (ANMs) transporting particles from the north to Penghu and Dongsha Islands. The most abundant water-soluble ionic species in PM2.5 were SO42−, NO3− and NH4+, catalogued as the secondary inorganic aerosols (SIAs). Meanwhile, the most abundant metals in PM2.5 were crustal elements (Mg, K, Ca, Fe, and Al), while the concentrations of trace metals (V, Cr, Mn, Ni, As, Cd, and Pb), mainly from anthropogenic sources, also increased from fall onwards. Organic carbon (OC) was the main species of carbonaceous content of PM2.5 in all seasons, and OC/EC ratios increased during the seasons with prevailing northeastern winds. Anhydrosugar, concentrations in winter and spring were generally higher than those in summer and fall, indicating significant biomass burning occurring in winter and spring. Correlation analysis showed a high correlation between PM2.5 concentrations and chemical composition between two subtropical islands. The correlation of chemical composition for different transportation routes revealed that northern routes had a higher correction than southern routes. Overall, the cross-boundary PM, accounted for 28.4–61.0% and 36.4–76.8% at Penghu and Dongsha Islands, respectively, significantly impacted local air quality, particularly at Dongsha Islands.