In recent decades, innovations in both direct interface (e.g. touchscreen based systems) and indirect interface (e.g. remote controller based systems) have successfully entered consumer markets. These solutions became major channels of infotainment function interaction. However, the popularity of new Human-Machine-Interfaces (HMIs) also comes with growing concerns for driver distraction. It is not a trivial quest to design a system that can make functions accessible to drivers as well as maintain drivers’ ability to cope with the complex driving task. To understand driver distraction, eye behavior has been studied extensively with a focus on off-road glances. Several standards and guidelines are based on off-road glance-related measures. An alternative approach is to consider both on-road and off-road glances. This can be done using an algorithm such as Kircher and Ahlstrom’s (2009) AttenD algorithm. This type of approach offers unique insight into understanding HMI interactions -- and focuses on whether drivers can maintain their attention to the roadway. Studies applying AttenD have revealed significant relationships between attention allocation and crash potential (Seppelt et al., under review) as well as correlations with established workload surrogate measures (Lee et al. under review). Inspired by these applied AttenD studies, we compared two prevalent visual-manual HMIs. The two HMIs of interest are a touchscreen-based interface (a production system) and a remote rotary controller based interface (a high-fidelity prototype). Five typical in-vehicle controlling tasks were evaluated, including a continuous-control task, a shortcut task, a menu-navigation task, a list-operation task and a function-switch task. In addition, a radio task was used as the reference. Sixteen participants’ glance behavior during each task were recorded and manually coded to apply the AttenD algorithm. Results suggested that with a higher-positioned display and haptic feedback, the remote rotary controller helped drivers maintain attention to the roadway better than a touchscreen-based interface for simple continuous control and shortcuts tasks. For the more complex tasks, the results are mixed with interesting insights. Additionally, the AttenD also revealed significant individual differences in attention management strategy. In summary, an algorithm such as AttenD not only can compare different HMIs, but also can reveal driver attention allocation strategy.