Reaction time is a surrogate measure of alertness and fatigue, which is essential to brain-computer interface (BCI) research in understanding and preventing fatigue in driving. Accurate reaction time determination is crucial, but complicated by the continuous nature of steering wheel inputs. Continuous data requires the selection of thresholds to determine valid reactions, and such thresholds may be under-sensitive or over-sensitive. In this study, steering wheel and electromyography (EMG) data collected from a BCI driving fatigue experiment was analyzed to determine both mechanical movement onset and EMG activation times. Several automatic methods for detecting wheel movement onset were evaluated against manual detection by experimenters. Results showed that single threshold detection with output smoothing yielded nearly identical results, whilst allowing for automatic determination of erroneous trials. EMG data analysis also showed that activation of the triceps long head muscles consistently preceded mechanical movement onset by 0.1s. These results collectively demonstrate that steering reaction time can be accurately determined through automatic methods, and that EMG activation data can be used as verification or a substitute for mechanically determined reaction times, thereby laying the groundwork for further BCI research into driving fatigue.