This paper deals with the results of the fatigue tests on four kinds of structural steels for ships, including ordinary mild steels and high strength steels. Main objective of this research is to investigate the influence of notch and welding on the fatigue strength of the steels. The following conclusions were obtained from the experimental results. 1) In the case of pulsating tension tests, the S-N curves converge approximately toward the tensile strength of steels (almost parallel to the N-axis at stresses higher than 90 per cent of the tensile strength), where no significant difference was found in the fatigue strength between notched or welded specimens and plain specimens. In general, the low cycle fatigue strength of steels is governed by their tensile strength. 2) When the maximum stress is lower than 90 per cent of the tensile strength of steel, either notched or welded specimens showed lower strength than the plain specimens. It was also found that the significant decrease of the fatigue strength in highly stress-concentrated specimens appeared in comparatively shorter life than that of less stress-concentrated specimens. 3) The fatigue strength of plain specimens is approximately proportional to the tensile strength of the material. 4) In the case of notched specimens, the fatigue strength for a large number of cycles is not proportional to the tensile strength of steels. The difference of the endurance limit between mild steels and high steels is not so pronounced as in the case of plain specimens. This tendency becomes remarkable in sharply notched specimens. 5) Except in the case of heat-treated high strength steels, no significant influence of the metallurgical change on the fatigue strength of steels was observed, and therefore it is considered that the decrease in the fatigue strength of the welded specimens is mainly caused by stress concentration i. e. geometrical effect of the weld metals. On the other hand, welded joints of a heat-treated high strength steel showed considerable decrease of the fatigue strength as influenced by both metallurgical and geometrical defects due to welding. 6) Butt-welded joints by manual welding indicated higher fatigue strength than automatically welded specimens because of the difference in the form of the reinforcement.