Abstract. During August 2004 an Aerosol Time-of-Flight Mass Spectrometer (TSI ATOFMS Model 3800-100) and an Aerodyne Aerosol Mass Spectrometer (AMS) were deployed at Mace Head during the NAMBLEX campaign. Single particle data (size, positive and negative mass spectra) from the ATOFMS were imported into ART 2a, a neural network algorithm, which assigns individual particles to clusters on the basis of their mass spectral similarities. Results are very consistent with previous time consuming manual classifications (Dall'Osto et al., 2004). Three broad classes were found: sea-salt, dust and carbon-containing particles, with a number of sub-classes within each. The Aerodyne (AMS) instrument was also used during NAMBLEX, providing online, real time measurements of the mass of non-refractory components of aerosol particles as function of their size. The ATOFMS detected a type of particle not identified in our earlier analysis, with a strong signal at m/z 24, likely due to magnesium. This type of particle was detected during the same periods as pure unreacted sea salt particles and is thought to be biogenic, originating from the sea surface. AMS data are consistent with this interpretation, showing an additional organic peak in the corresponding size range at times when the Mg-rich particles are detected. The work shows the ATOFMS and AMS to be largely complementary, and to provide a powerful instrumental combination in studies of atmospheric chemistry.