Many rotating machinery components fail due to fatigue when subjected to continuous fluctuating stresses. Hence, estimation of fatigue crack initiation life is essential to avoid catastrophic failure. Effective vibration based fatigue life analysis requires measurement of accurate time varying signal. In this study, experimentally observed fatigue lives of rotating shaft, for three different notch configurations, are compared with fatigue lives estimated using two approaches based on an acquired vibration signal. The first one is time domain approach (based on Rainflow cycle counting) while the second one is frequency domain approach (based on power spectral density moments). In the frequency domain approach, fatigue life is estimated using the narrow-band approximation and Dirlik’s empirical solution. The performance of two approaches in estimating fatigue life for the same signal length taken at different time intervals from the total signal acquired is also discussed. In addition, experimental uncertainty analysis is performed and discussed in this study. A good correlation is found between the estimated fatigue life using Dirlik’s rainflow range probability density function and experimental life. Therefore, this study concludes that the Dirlik’s approach can be considered as preferable method for estimating fatigue life of rotating shaft. © 2018, The Korean Society of Mechanical Engineers and Springer-Verlag GmbH Germany, part of Springer Nature.