The present study investigated the effect of aging on direct navigation measures and self-reported ones according to the real-virtual test manipulation. Navigation (wayfinding tasks) and spatial memory (paper-pencil tasks) performances, obtained either in real-world or in virtual-laboratory test conditions, were compared between young (n = 32) and older (n = 32) adults who had self-rated their everyday navigation behavior (SBSOD scale). Real age-related differences were observed in navigation tasks as well as in paper-pencil tasks, which investigated spatial learning relative to the distinction between survey-route knowledge. The manipulation of test conditions (real vs. virtual) did not change these age-related differences, which are mostly explained by age-related decline in both spatial abilities and executive functioning (measured with neuropsychological tests). In contrast, elderly adults did not differ from young adults in their self-reporting relative to everyday navigation, suggesting some underestimation of navigation difficulties by elderly adults. Also, spatial abilities in young participants had a mediating effect on the relations between actual and self-reported navigation performance, but not for older participants. So, it is assumed that the older adults carried out the navigation task with fewer available spatial abilities compared to young adults, resulting in inaccurate self-estimates.