BackgroundThis article addresses how to combine three elements (a pictogram, an arrow, a city) in a variable message sign (VMS) to locate temporary events (e.g., “congestion before Milan”). We adopted the G1c stack model as a design template, an Advanced Directional Sign (ADS) recommended by the 1968 Convention to locate cities, which can be easily adapted to modern VMS. However, as most of the VMS in operation are not full-matrix, we have also adapted this design to more restrictive display conditions. This adaptation critically concerned the arrow function on the message that either points up broadly (generically, as in G1c) or connects with the city more specifically (explicit). Although G1c reads top-down like a verbal text, previous studies indicated drivers’ preference for bottom-up landmark order in VMS, so both ordering criteria were compared in the present study. MethodsThe experiment involved 99 people (70 drivers and 29 drivers in training). Participants were informed that they would see various VMS reporting certain events (e.g., congestion) related to one of four cities along the road. Their task was to identify the event location (before, after the city) after seeing blocks of two consecutive messages (first a complementary message, then the target message), limiting their response to the content of the second message. Three design-focused factors were tested: typographical alignment (left or centre), landmark order (bottom-up or top-down), and arrow function (explicit or generic). The rate of correct location answers was the dependent variable. ResultsResults revealed that comprehension varied greatly depending on the arrow’s function and the placing of elements. In the explicit-arrow messages, comprehension was good both in the Top-down and Bottom-up conditions, but in the generic-arrow messages, only in the Bottom-up condition was comprehension good. Likewise, understanding was better in the Before condition than in the After condition in all combinations of Landmark order and Arrow function conditions. In general, left alignment of the central column elements of the VMS improved comprehension respective to centred alignment. Finally, the complementary message factor had an effect under certain circumstances. Practical implicationsThe messages displaying a generic arrow (following the G1c model) were better understood when the landmarks were ordered bottom-up, not top-down. In addition, explicit-arrow messages were better understood per se (in the absence of a complementary message) than generic-arrow messages. Overall, this work suggests that improving our understanding of how thought processes and design features relate to each other can contribute to safer driving nationally and internationally.