This study investigates the long-term variability and evolution of particulate matter with an aerodynamic diameter of <2.5 µm (PM_2.5) sources in Vilnius, Lithuania, during the period 2013–2021. Source apportionment was performed using Positive Matrix Factorization (PMF) based on elemental composition data obtained through particle-induced X-ray emission (PIXE) analysis. The results revealed substantial year-to-year variability in the chemical profiles of the identified sources. Crustal/mineral dust was characterized by high contributions of lithogenic elements, including Si, Ca, Ti, and Fe, while soil dust exhibited elevated proportions of Al, Ca, and Fe. Traffic non-exhaust emissions were marked by elevated Cu, Zn, and Pb in 2013–2015, whereas exhaust emissions in 2019–2021 were characterized by sulfur-rich aerosols. Industrial and oil combustion sources showed enhanced contributions of Ni, V, and Cr, particularly in 2016, 2018, and 2020. Biomass/wood burning represented a major seasonal source, reaching peak intensity in 2018–2019 and characterized by elevated K and Zn contributions. A notable long-term trend was the increasing importance of soil-derived particles, as reflected by Al contributions rising to 91.2% by 2021. Overall, the major PM_2.5 source categories remained relatively stable, while their chemical fingerprints and relative importance exhibited substantial temporal variability.