Traffic light assistance systems (TLASs) can be infrastructure based or on-board, and in the latter case they can inform the driver about the time remaining to green, or about the recommended speed for a smooth passage at green. A speed-based and a time-based on-board system prototype was compared against each other and against a baseline without any assistance system. Using a within-subjects design, 18 participants drove in a fixed-base simulator along a suburban road with signalised intersections, where the delay to green was set to zero (allowing a passage at the current speed), “half-speed” (requiring a clear speed reduction) and “stop” (requiring a substantial speed reduction). Driving behaviour, visual attention distribution and acceptance were evaluated. Both support systems improved driving efficiency and comfort over baseline, with the time-based system achieving higher scores in general. Both systems attracted a substantial amount of visual attention in the current setting; however, single-glance durations were below 1 s, and the number of glances forward were equal in the time-based condition compared to baseline, but lower in the speed-based condition. No red or amber light violations were registered in baseline, while some occurred with any of the systems. Acceptance for both systems was high, with higher scores for the time-based prototype. Overall, an on-board TLAS with a countdown timer to green has the potential to increase efficiency and comfort without strong indications for attention disruption, but the risk for increased red/amber light violations has to be addressed. Improved system design as a way to mitigate potential issues is discussed.