Fatigue caused by forced vibration of a random nature is one of the major concerns in the automotive field. Random loading of components under actual driving conditions causes dynamic stress/strain responses which can be better described and handled in the frequency domain. Power Spectrum Density (PSD) is usually the most concise and straightforward way of representing a random process. Since frequency domain methodologies are gaining more respect and interest it is very important to be aware of their limitations and scope, particularly when compared to time domain algorithms. The present paper aims to discuss both approaches and establish some comparisons in terms of accuracy, range of application, computation time and user friendliness.