Previous behavioural research (Land and Horwood in Nature 377:339-340, 1995) indicates that surprisingly little visual information is required to effect smooth and accurate steering through a curving roadway. Based on results from a driving simulator study, Land and Horwood reported that viewing the roadway through two horizontal apertures, one degree of visual angle in height, can result in steering performance which is indistinguishable from that obtained with the whole scene visible. The position of the apertures in the visual field, they claimed, is crucial; higher up leads to less accurate but more stable steering responses, whilst lower down leads to jerky steering but more accurate lane-keeping. These findings are consistent with a two-stage model of steering control proposed by Donges in Human Factors 20:691-707 (1978). However, in a driving simulator, the temporal lag between the input signal received from a control device and the output presented on the display can have profound effects on steering behaviour. In two experiments, we show that the effect of aperture position on steering accuracy and stability is pronounced when a slow frame update rate is used (7.2 Hz, as used by Land and Horwood in Nature 377:339-340, 1995), but largely attenuated with a faster update rate (72 Hz). These results are consistent with the broader empirical literature dealing with temporal delays in manual tracking, and urge a critical reappraisal of the behavioural evidence for the two-stage steering model.