Single airguns and airgun arrays of different size and volume are used as sound sources for scientific marine seismic reflection and refraction surveys conducted by R/V Polarstern in the Arctic and Antarctic Ocean. To ensure that these research activities do not affect marine wildlife and particularly marine mammals in the Antarctic Treaty Area south of 60°S knowledge of the sound pressure field of the seismic sources is essential. Therefore, a broadband marine seismic source characterization study was conducted at the Heggernes Acoustic Range in the Herdlefjord, Norway in October 2003. The objectives were (1) to determine the spatial distribution of the sound pressure levels emitted by Polarstern's seismic sources, (2) to compute the source levels assuming a spherical amplitude decay, (3) to determine mitigation radii, within which at least some species of marine mammals might possibly experience behavioral or physiological disturbance due to the received sound pressure levels. The thresholds currently in use to determine mitigation radii are 160 dB(RMS) re 1 µPa for potential behavioral disturbance and 180 dB(RMS) re 1 µPa for potential physiological and hearing effects like temporary threshold shifts.To determine the spatial distribution of the sound pressure levels each airgun (array) was shot along a line of 2-3 km length running between 2 hydrophone chains with receivers in 35, 100, 198 and 263 m depth. A GI-Gun (2.4 l), a G-Gun (8.5 l) and a Bolt PAR CT800 (32.8 l) were deployed as single sources, and 3 GI-Guns (7.4 l), 3 G-Guns (25.6 l) and 8 VLF-Guns (24 l) as arrays. The measurements are complemented by a modeling approach for an 8 G-Gun (68.2 l) and 8 G-Gun+1 Bolt PAR CT800 array (100.1 l). The data analysis includes a determination of peak-peak, zero-peak and RMS-amplitudes, sound exposure levels and amplitude spectra as function of source-receiver distance.The amplitude vs distance graphs, analyzed for the 4 hydrophone depths, show the typical directivity of marine seismic sources. Due to the destructive interference of the direct wave and the ghost reflection, amplitudes almost vanish close to the sea surface and are maximum in several hundred meters depth. A comparison between the amplitudes recorded during approach and departure reveals a shadowing effect of Polarsterns's hull. Amplitudes recorded at the same source-receiver distance are lower during approach than during departure indicating that the ship's hull deflects sound propagation forward the ship. Mitigation radii derived from the amplitude vs distance graphs of the deepest hydrophone for the 180 dB(RMS) level vary between 200-600 m. Extrapolated source levels range from 224-240 dB(0-pk) re 1 µPa @ 1 m. Spectral peak levels occur below 100 Hz, amount to 182-194 dB re 1 µPa/Hz @ 1 m and decrease by ~30 dB re 1 µPa/Hz within the 1 kHz range. A first modeling approach of source directivities, based on the assumptions of deep water, a homogeneous water column and a linear superposition of the single airgun signatures of an array, shows a good agreement between the amplitude decay curves of the G-Gun signals recorded at the 2 deepest hydrophones but some discrepancies for the recordings at the 2 shallow hydrophones. Further studies are required here.