Metal workshops are workplaces with the substantial production of particulate matter (PM) with high metal content, which poses a significant health risk to workers. The PM produced by different metal processing techniques differs considerably in its elemental composition and size distribution and therefore poses different health risks. In some previous studies, the pollution sources were isolated under controlled conditions, while, in this study, we present a valuable alternative to characterize the pollution sources that can be applied to real working environments. Fine PM was sampled in five units (partially specializing in different techniques) of the same workshop. A total of 53 samples were collected with a temporal resolution of 30 min and 1 h. The mass concentrations were determined gravimetrically, and the elemental analysis, in which the concentrations of 14 elements were determined, was carried out using the X-ray fluorescence technique. Five sources of pollution were identified: background, steel grinding, metal active gas welding, tungsten inert gas welding, and machining. The sources were identified by positive matrix factorization, a statistical method for source apportionment. The identified sources corresponded well with the work activities in the workshop and with the actual sources described in previous studies. It is shown that positive matrix factorization can be a valuable tool for the identification and characterization of indoor sources.