In driver-focused vehicle development, driver workload is generally evaluated subjectively, with physiological, psychological, and behavioral indexes used to quantify and substantiate the subjective rating.In contrast, a model of driver behavior expresses the driver’s behavioral characteristics which make it possible to estimate how the driver will incorporate information into vehicle operation. Therefore, it is presumed to be capable of estimating the internal state of a driver. Conventionally, a model of driver behavior related to pedal operation has been used for evaluating the driver’s habits and the acceptability of various types of support devices. However, it has not been used for estimating driver workload related to pedal operation.To examine driver workload, this study divided pedal operation magnitude into two components: a learning/judgment component and a correcting component for prediction errors. A method was devised of separating these two components. Furthermore, an operation model was developed for the correcting component, which is considered strongly related to driver workload. Finally, the study clearly identified the relationship between the visually perceived information and pedal operation magnitude.Three test vehicles were used that provided subjectively different levels of driver workload. Based on the driving experiment results, an operation model was used to identify the relationship between the visually perceived information and pedal operation magnitude as a model parameter. Results indicated that this model parameter could be used to compare driver workload.