A survey has been made of the developments of welding processes and the literature concerned with the fatigue resistance of butt welded joints. Evident from the literature survey was the fact that a mass of experimental data was available on the fatigue properties of welds yet many fatigue failures were still being reported. It appears essential therefore that the design data derived from experimental results should be presented in a meaningful way to aid the safe design of welded connections. Fatigue tests were conducted on specimens cut from 1/2 in. thick plate of medium carbon steel, over a range of alternating and mean stress levels. The welds were produced using manual metal arc, gas metal arc and electron beam welding processes and manufactured in industry by accepted production methods. A non destructive examination together with static tests were carried out on both the parent material and the welded joints. The results obtained have enabled a comparison to be made between the data obtained from static, dynamic and non destructive tests. The analysis shows some evidence of correlation between the information derived from the static and non destructive tests with that of fatigue performance. A method of analysing fatigue data, based on the method of least squares, has been developed. This provides good fits to data when used in conjunction with modified versions by Goodman and Gerber of the basic Jefferson empirical fatigue equation. The fatigue curves generated by this method can be described by two equation constants which can subsequently be used to generate a family of fatigue curves over a range of mean loads. The method of analysis is also capable of calculating confidence limits on experimental data for both stress and fatigue life for the predicted fatigue curve.