The paper gives an overview of techniques used for characterization of IC-engines as acoustic sources of exhaust and intake system noise. Some recent advances are introduced and discussed. Linear frequency domain prediction codes are frequently used for calculation of low frequency sound transmission in and sound radiation from IC-engine exhaust and intake systems, even though nonlinear time domain models are also developing fast. To calculate insertion loss of mufflers or the level of radiated sound information about the engine as an acoustic source is needed. The source model used in the low frequency plane wave range is often the linear time invariant one-port model. The acoustic source data is obtained from experimental tests or from 1-D CFD codes describing the engine gas exchange process. Multi-load methods and especially the two-load method are most commonly used to extract the source data. The IC-engine is a high level acoustic source and in most cases not completely linear. It is therefore of interest to have models taking weak non-linearity into account while still maintaining a simple method for interfacing the source model with a linear frequency domain model for the attached exhaust or intake system. Some years ago a model which can consider weakly non-linear sources was presented, which gave an improvement over the traditional two-load technique for determining source data from experiments. It is however fairly complicated to implement and has not been used a lot. In this paper an alternative technique based on so called polyharmonic distortion modeling, used for nonlinear characterization of microwave systems is introduced and tested. Comparisons are made with the results from linear source models and the previously published weakly nonlinear source model.