Abstract For the exploration of the dominant aerodynamic noise sources, the pressure fluctuations in the exit plane (near field) of the propfan model CRISP (Counter Rotating Integrated Shrouded Propfan) were measured with conventional 1/4 inch microphones. The pressure field of the tone components was resolved into a distribution of duct modes. Knowledge of the dominant modes allows conclusions about the dominant noise generation mechanisms, because the different noise sources inside the propfan create different sets of modes. The CRISP concept is developed by Motoren- und Turbinen-Union München (MTU). The experimental propfan has two counter-rotating rotors of 0·4 m diameter and equal speed. The shroud is supported by seven struts located downstream of the second rotor. Measurements were made with equal (B1=B2=10) as well as unequal (B1/B2=10/12) blade numbers and under different operational conditions. The highest overall harmonic levels were found for the configuration with equal blade numbers. In this case, the blade passing frequency component is generated mainly by the rotor 2/struts interaction, and the higher blade tone harmonics, which dominate the overall tone noise level, are produced by the interaction of the two rotors. In the case of unequal blade numbers, all even harmonics of the shaft frequency (H=2, 4, 6 ...) can be generated by the rotor 1/rotor 2 interaction. The harmonics belowH=22, however, are excited as non-propagational modes only and were found to have small amplitudes in the exit plane. The rotor 1/rotor 2 interaction is the main noise generation mechanism for the configuration with a short axial distance between the rotors. When the rotor distance is enlarged, the rotor/rotor interaction noise is reduced and, as a consequence, the contributions from the rotor/struts interactions become important. In addition to the experiments, a theoretical method is described for the prediction of the frequencies and azimuthal modes generated by two rotors with arbitrary speeds, directions of rotation and blade numbers. This method is helpful for the design of low noise rotor systems.