Number size distributions of atmospheric aerosol particles were simultaneously measured at a kerbside and an urban background site in the city of Thessaloniki, Greece, from June to October 2009. New particle formation events were observed ca. 27% of the days at the urban kerbside site and 29% of the days at the urban background site. In almost all the cases the events started between 10:00 and 12:00, and continued for several hours. The total number concentration (TNC) of the particles having diameters from 10 to ca. 500 nm during the events increased from 1.4 × 10⁴ to 6.5 × 10⁴ #/cm³ at the urban kerbside site, and from 0.2 × 10⁴ to 2.4 × 10⁴ #/cm³ at the urban background site. At the urban kerbside site, 9% of the days exhibited class I events (i.e., events followed by a clear growth of the newly formed particles), 10% class II (i.e., events during which the concentration of nucleation mode particles were high but their growth was not continuous), 67% were characterised as non-event days, and 14% of the days exhibited no clear particle formation pattern (undefined). At the urban background site, 15% of the days were classified as class I, 5% as class II, 75% of the days showed no nucleation, whereas only 5% of the days were undefined. While the fraction of event days (both class I and class II) at both sites was similar (ca. 20%), the higher fraction of class I events observed at Eptapyrgio can be attributed to the cleaner environment of the urban background site that allows better identification of the particle concentration increase. The nucleation bursts show a similar pattern at both sites, with the newly formed particles reaching a final size of ca. 80-100 nm. A distinct difference between the two stations was that the smallest particles observed during the new-particle formation events had a diameter of ca. 10 nm (i.e., the smallest particles we could observe) at the kerbside site and ca. 20 nm at the urban background site. This is an indication that the new particles observed at the urban background station are formed elsewhere and are transported to the site. Estimated concentrations of H₂SO₄ using a proxy model, suggest that these are high enough to explain the nucleation events despite that the available aerosol surface was high, especially at the urban kerbside site.