Most fatigue loaded structural components are subjected to variable amplitude loads which must be processed into a form that is compatible with design life calculations. Rainflow counting allows individual stress cycles to be identified where they form a closed stress–strain hysteresis loop within a random signal, but inevitably leaves a residue of open data points which must be post-processed. Comparison is made between conventional methods of processing the residue data points, which may be non-conservative, and a more versatile method, presented by Amzallag et al. (1994), which allows transition cycles to be processed accurately.This paper presents an analytical proof of the method presented by Amzallag et al. The impact of residue processing on fatigue calculations is demonstrated through the application and comparison of the different techniques in two case studies using long term, high resolution data sets. The most significance is found when the load process results in a slowly varying mean stress which is not fully accounted for by traditional Rainflow counting methods