Previous research has shown age-related declines in the use of specifying perceptual information to guide action decisions in traffic environments. In the present study, we investigated the effects of cross-modal cueing on perceptual training with older adults in a virtual road-crossing task. Specifically, we tested whether the visual information used to decide which inter-car gaps afforded crossing could be influenced by sound events which tracked either gap-specifying or non-specifying optic variables. Thirty-nine older adults were divided into three groups who practiced with auditory cues mapped to either the time-to-arrival of the approaching car (specifying group), its distance (non-specifying group), or no sounds (control group). Although all three groups reduced decision errors with training, analysis of which variables predicted crossing responses showed that the specifying group’s decisions became more attuned to the time-to-arrival information, whereas the non-specifying group became less attuned to this information and more to the distance information. Thus, attention for action decisions in older adults was re-educated towards either specifying or non-specifying visual information, depending on the optic variables highlighted by the auditory cues. The implications of these findings are discussed in the context of designing perceptual learning studies and road safety interventions for the elderly.