Increasing demand for the extended lifetime of structures stresses the need for reliable high-cycle fatigue lifetime predictions for variable amplitude-loaded structures. Design standards (including Eurocode 3, AASHTO and British Standard 7608) provide a fatigue design model based on constant amplitude S-N curves and a linear damage accumulation model. There are minor differences in detail categories between standards, but larger differences between the extensions of the S-N curves for variable amplitude loading. It is known that the load spectrum influences this extension due to load interaction and sequence effects. The validity of the predicted fatigue life by available design standards is checked using a compilation of variable amplitude fatigue test data for steel arc-welded joints from the literature with different loading spectra. The results indicate that the standards are generally conservative in the mid- but not in the high-cycle regime, suggesting that resistance non-linearities should be explicitly addressed.