Studies on the mismatch between objective and perceived measures of walkability and walking provide insights into targeting interventions. These studies focused on those living in more walkable environments, but perceiving them as less walkable. However, it is equally important to understand how the other mismatch (living in less walkable areas, but perceiving them as walkable) is related to walking. This study examined how the mismatch between perceived and objective walkability measures (i.e., living in less walkable areas, but perceiving them as walkable, and living in more walkable areas, but perceiving them as less walkable) was associated with walking. Baseline data from adult participants (n = 1466) of the RESIDential Environment Project (Perth, Australia in 2004-06) collected self-report neighborhood walking for recreation and transport in a usual week and participants' perceptions of street connectivity and land use mix in their neighborhood. The exposure was the mismatch between objective and perceived measures of these. Multilevel logistic regression examined associations of walking with the mismatch between perceived and objective walkability measures. Perceiving high walkable attributes as low walkable was associated with lower levels of walking, while perceiving a low walkable attribute as walkable was associated with higher levels of walking. Walking interventions must create more pedestrian-friendly environments as well as target residents' perceptions.