Solid round bars with a shoulder fillet are frequently used in engineering structures because it facilitates a smooth transition in thickness. However, the presence of shoulder fillets is currently neglected in the prediction of fatigue crack growth in engineering practice. Since stress concentrating locations like a shoulder fillet can have a significant effect on the fatigue crack growth behaviour, neglecting the presence of a shoulder fillet may introduce unacceptable inaccuracies in the current fatigue crack growth predictions done in solid round bars with a shoulder fillet. Experimental research was done on characterizing the effect of shoulder fillets on the fatigue crack growth behaviour in solid round bars loaded in tension. The problem was approached from two points of view: the conventional stress intensity approach and a novel energy approach. Proven methodologies were used for the stress intensity approach, the energy approach had a more exploratory character. The first goal of the research was to investigate the influence of a shoulder fillet on the fatigue crack growth behaviour in solid round bars loaded in tension. The influence of a shoulder fillet was quantified by the development of a stress intensity solution based on the results of fatigue crack growth experiments. The developed stress intensity factor solution was implemented in a fatigue crack growth prediction model for solid round bars with a shoulder fillet loaded in tension. The stress intensity factor range is conventionally used as similarity parameter to predict fatigue crack growth. An issue with this similarity parameter that has received a lot of attention in the past decades is the effect of the stress ratio. Empirically derived equations are proposed in literature to take the effect of the stress ratio into account for fatigue crack growth predictions. Recent literature suggests that this effect of the stress ratio is a reflection of the inadequacy of the stress intensity factor range to uniquely describe the applied load cycle; it was suggested that these stress ratio correction equations mostly compensate for the fact that cyclic stress has been used in a methodology where cyclic energy should have been used instead. The second goal of the research was to investigate how the energy used for a fatigue crack to grow in a complete load cycle relates to the fatigue crack growth rate. A prediction methodology was developed based on the energy released in a complete fatigue cycle rather than at maximum and/or minimum load as is the case in the conventional stress intensity approach. It was found that the average cyclic strain energy release rate could be used as similarity parameter in the developed methodology. The methodology was used to develop fatigue crack growth prediction model for solid round bars with a shoulder fillet loaded in tension. As a result of the research project, it was concluded that the developed fatigue crack growth prediction model, which makes use of the developed stress intensity factor solution, is able to predict fatigue crack growth for different shoulder fillet geometries at one stress ratio; it was less accurate at other stress ratios. It is recommended to do further research to validate the developed stress intensity factor solution. Next to that, it was concluded that the proposed prediction methodology with the average cyclic strain energy release rate as similarity parameter was able to predict fatigue crack growth independently of the applied stress ratio and shoulder fillet geometry. Points of further research are identified to continue the development of the average cyclic strain energy release rate as similarity parameter for the prediction of fatigue crack growth in general.