Analysis of chemical compositions of fine particle matters (PM2.5) in contaminated atmospheres has increasingly become a great concern of both scientific and technological importance because the contaminant concentrations of PM2.5 are playing an important role in affecting PM2.5-induced adverse effects on environment and human health. In this paper, ultrasound-assisted extraction coupled with ICP-MS was developed for a simultaneous quantification of 12 toxic elements (Be, Al, Cr, Mn, Ni, As, Se, Sb, Tl, Pb, Hg and Cd) in PM2.5 samples. The proposed method was validated with several parameters, namely linearity of calibration, precision and accuracy, limit of detection (LOD) and limit of quantification (LOQ). The accuracy of the proposed method was validated with recovery tests of different samples spiked with corresponding elements at known concentrations (5, 10 and 20 μg/L) and the average percent recovery varied in the range of 87.3–101.3%. The results showed that this method was a multi-element technique featuring high precision and accuracy as well as low LOD. This proposed method was applied to a simultaneous determination of the 12 target elements in atmospheric fine particles collected from a heavily polluted urban residential area in China, and showed that the particles were rich in Al, Pb and Mn, followed by As, Sb, Cr, Cd, Se and Ni. The main advantages of the ultrasound-assisted extraction process were fast sample treatment and low consumption of reagents in the extraction process.