Simulation tools usage in the automotive industry has faced a significant increase in the last decade, due to the significant advantages of time and cost reduction opportunities it can potentially provide. Therefore new modeling techniques are constantly developed in order to increase the correlation between simulated and physical test results. In particular, modeling of spot welds plays a very important role in defining the accuracy of a finite element model, mainly with large models, such as a complete trimmed body, have to be dealt with. The properties of a welded set depends on a significant number of variables, such as the temperature during welding, the compression load created by the welding probe and geometric variation of the nugget position. Although it is not possible to take all those variables into account in a finite elements model, a good modeling strategy should produce acceptable results even when those are neglected or only modeled in an approximate way. This work compares three spot welds modeling strategy with respect to their performance on a normal vibration modes analysis. The first strategy consists of representing the spot-weld as a common node between the bonded shells; the second one uses rigid elements as connection and in the third one the weld nugget is modeled as a solid element. The results for the three techniques are discussed and compared with physical testing in order to identify the more accurate one for normal modes analysis.