Driver fatigue is a risk factor for road crashes. Fit for duty technologies could play a pivotal role in countering these crashes. Heart rate variability (HRV) and the pulse wave shape are influenced by the autonomic nervous system and are therefore affected by fatigue. This work focusses on modelling their relationship with fatigue and is based on data recorded in a simulated driving study. Six different multivariate linear regression models, using either stepwise variable selection or principal component analysis, are presented in this study. To account for differences in physiology, individual participant baselines for HRV and pulse wave parameters are introduced. Stepwise regression using any kind of baseline yields the most promising results. The most promising predictors are the ratio LFHF between low and high frequency components of HRV and heart rate. Finally, a stepwise regression model with a baseline, which has an adjusted R2 statistic of 0.17, is proposed for further use. Nevertheless, further research with an extended dataset is necessary, incorporating a more diverse participant group and a higher number of recordings from severely sleepy drivers.