In fall 2014 non – methane volatile organic compound (NMVOC) concentrations were measured with a PTR – QiTOF mass spectrometer at the University of Innsbruck (Austria) over a period of 55 days. 138 NMVOCs were used to perform non–negative matrix factorization (NMF). The Goal was to find out whether the resulting factors could be assigned to chemical sources, and more specifically whether biogenic and anthropogenic sources can be separated by the method. Ancillary data from an automatic weather station near the sampling site, as well as a foehn Event classification were used to identify potential sources depending on meteorological constraints. It was determined that six NMF factors could describe the variability of the entire NMVOC dataset reasonably well. It was determined that six NMF factors could describe the variability of the entire NMVOC dataset reasonably well. It was possible to identify very localized sources (e.g. venting air outlets from laboratories within the building). Meteorological data, especially days with foehn proved useful to allocate a factor to the clinic area south of the university building. However, biogenic and anthropogenic sources could not be separated completely, owing to the similar diurnal variations of some biogenic and traffic related emissions.