[en] ABSTRACT: This study investigated seasonal variations in PM_1_0 and PM_2_._5 mass and associated trace metal concentrations in a residential area in proximity to the crude oil refinery plants and industrial parks of central Taiwan. Particle measurements were conducted during winter, spring and summer in 2013 and 2014. Twenty-six trace metals in PM_1_0 and PM_2_._5 were analyzed using ICP-MS. Multiple approaches of the backward trajectory model, enrichment factor (EF), Lanthanum enrichment and positive matrix fraction (PMF) were used to identify potential sources of particulate metals. Mean concentrations of PM_1_0 in winter, spring and summer were 76.4 ± 22.6, 33.2 ± 9.9 and 37.4 ± 17.0 μg m"−"3, respectively, while mean levels of PM_2_._5 in winter, spring and summer were 47.8 ± 20.0, 23.9 ± 11.2 and 16.3 ± 8.2 μg m"−"3, respectively. The concentrations of carcinogenic metals (Ni, As and adjusted Cr(VI)) in PM_1_0 and PM_2_._5 exceeded the guideline limits published by WHO. The result of EF analysis confirmed that Mo, Sb, Cd, Zn, Mg, Cr, As, Pb, Cu, Ni and V were attributable to anthropogenic emission. PMF analysis demonstrated that trace metals in PM_1_0 and PM_2_._5 were from the similar sources, such as coal combustion, oil combustion and traffic-related emission, except for soil dust and crustal element emissions only observed in PM_1_0 and secondary aluminum smelter only observed in PM_2_._5. Considering health-related particulate metals, the traffic-related emission and coal combustion for PM_1_0 and PM_2_._5, respectively, are important to control for reducing potential carcinogenic risk. The results could aid efforts to clarify the impact of source-specific origins on human health. - Highlights: • Multiple approaches to identify sources of PM_1_0 and PM_2_._5 metals were used. • Four similar sources contributed to metals in PM_1_0 and PM_2_._5 in the study area. • Six toxic metals fell below the limits of WHO Air quality guideline. • Control strategies can be proposed based on reducing potential carcinogenic risk.