The chemical composition of ambient aerosol particles affects numerous important aerosol parameters such as their hygroscopicity, optics, and mass as well as their potentially adverse health effects. The objective of this study was to derive both detailed chemical speciation and useful proxies for the quantitative classification of the organic matter (OM) content of carbonaceous aerosol samples. Using three different thermal desorption techniques in an inert atmosphere we investigated eight different carbonaceous particulate matter (PM) samples used for health effect studies: thermal desorption gas chromatography with mass spectrometry, evolved gas analysis with mass spectrometry, and thermogravimetry with Fourier transform infrared spectroscopy. The samples include different types of laboratory-generated particles (pigment black, diffusion flame soot, spark-generated carbon) and two ambient aerosol samples (diesel soot and particulates collected in a road tunnel). All samples showed increasing mass desorption with rising temperature, but no reliable OM classification was possible based on thermal mass desorption alone. In fact, the "organic-free" spark-generated carbon particles showed the second highest mass desorption at 800 degrees C due to the formation of oxygenated structures on unsaturated surface sites and the subsequent evolution of CO and CO2 at elevated temperatures. A quantitative OM classification was accomplished by combining measurements of thermogravimetry and mass spectrometry (up to 800 degrees C) into a novel parameter, the "apparent organic mass fraction". The validity of this classification was confirmed with a second proxy parameter, based only on the evolution of organic components during thermal desorption and information on the generation process of the particles. Both types of pigment blacks (Printex) samples and the spark-generated carbon particles showed the lowest apparent organic mass fraction (< 5%), whereas for road tunnel and diesel emission particles < 16 and < 19% was estimated, respectively.