The degree to which someone is mentally preoccupied by a traffic scenario depends on factors such as traffic density, roadway arrangement, number of different participants, and the amount of distracting objects. It can be assumed that with increasing complexity of these aspects, the mental workload of the driver is growing. In a series of studies traffic scenarios were characterized by measuring their structural complexity. These complexity measures were compared to the experienced mental workload of the car driver. To investigate these relationships in detail from the driver's perspective, videos of different dynamic traffic situations were recorded. These videos were then analysed by calculating the structural complexity of each video‐frame on the basis of a luminance‐change algorithm. In a further step these videos were shown to participants who had to judge the mental demands relating to the different traffic situations. The judgments were continuously registered by a hand‐held potentiometer so that a time‐series of mental workload values resulted. Correlation analyses between the time‐series of mental workload and the numeric measures of complexity revealed substantial correlations. The results show that the judged mental workload related to certain traffic situation, and can be predicted to some extent by a formal complexity analysis of the traffic scene. Considerations are suggested relating to the extent to which continuous complexity measurements can be used to predict driver fatigue.