Sixty-two topsoil samples were collected within the third ring road of Xi'an City in Northwest China and analyzed by X-ray fluorescence spectrometry for the concentrations of As, Ba, Co, Cr, Cu, Mn, Ni, Pb, V, and Zn. The pollution levels of trace metals were assessed by pollution index (PI) and Nemerow pollution index (NPI). Meanwhile, the sources of trace metals were apportioned by receptor models, including positive matrix factorization (PMF), UNMIX, and principal component analysis-multiple linear regression (PCA-MLR). The average concentrations of the trace metals analyzed in the urban soil exceeded the corresponding soil element background values of Shaanxi Province, especially for Co, which was 2.38 times higher than the corresponding background value. The mean of PI was 2.38 for Co, reflecting a moderate pollution level, and ranged from 1.07 to 1.72 for other trace metals, presenting slight pollution levels. The NPI of trace metals varied between 1.20 and 3.50 with an average of 2.00, indicating that trace metals presented slight pollution in 62.90% of soil samples, moderate pollution in 30.65% of soil samples, and heavy pollution in 6.45% of soil samples, respectively. Three sources of trace metals apportioned by the three receptor models were mixed nature and anthropogenic source, traffic exhaust, and industrial emissions. The contributions of them were 38.58%, 32.72%, and 28.70% from the PMF, 65.36%, 17.76%, and 16.88% through the UNMIX and 49.16%, 38.90%, and 11.94% via the PCA-MLR, respectively. Meanwhile, the study results suggested that the combined usage of multiple receptor models is a good method to apportion the source compositions and contributions of trace metals in urban soil.