This study involves the development of a fmite element model of a hennetic reciprocating compressor for noise prediction applications. Inherent difficulties in developing the finite element model of a complicated structure are discussed and appropriate modeling strategies are evolved. The development of the complete compressor fmite element model is carried out in two stages modeling of the compressor housing and the assembly of components into the compressor assembly. The compressor housing is isolated for detailed modeling. Geometry complexity, secondary masses, spring mounts, lap-joint and manufacturing variations pose challenges in developing a reliable model. Frequent comparisons are made with experimental mobility scans to obtain insights into the actual behavior of the modeled structure. When possible, weaknesses are located in the finite element model and corrected. After sufficient revisions, 23 natural frequencies ( excluding the rigid body modes ) are found for the compressor housing in the low frequency range ( below 2000 Hz ) of analysis. Forced response calculations are also used to correlate the analytical model and test data, with a maximum of 5% disagreement for the 14 natural frequencies that could be correlated. Compressor assembly modeling involves detailed solid modeling of internal components for inertia properties, developing reduced-degrees-offreedom models of mounting springs and modeling of the shockloop. Finally, the components are assembled and the compressor assembly is solved for its natural frequencies by the component mode synthesis method. Eighty seven natural frequencies below 2000 Hz ( excluding the rigid body modes ) are found for the compressor assembly model. This model can be used to predict velocity responses on the surface of the housing, with the internally generated forces as excitations. Velocity response data are directly used in sound prediction.