Background: Structural and functional brain imaging methods have identified age-related changes in brain structures involved in gait control. This cross-sectional study aims to investigate gray matter networks associated with gait control in aging using structural covariance analysis. Methods: Walking speed were measured in 326 non-demented older community-dwellers (age 71.3±4.5; 41.7% female) under three different walking conditions: normal walking and two challenging tasks: motor (i.e.; fast speed) and an attention-demanding dual task (i.e.; backward counting). Results: Three main individual gray matter regions were positively correlated with walking speed (i.e.; slower walking speed was associated with lower brain volumes): right thalamus, right caudate nucleus and left middle frontal gyrus for normal walking, rapid walking and dual-task walking condition, respectively. The structural covariance analysis revealed that prefrontal regions were part of the networks associated with every walking condition; the right caudate was associated specifically with the hippocampus, amygdala and insula for the rapid walking condition and the left middle frontal gyrus with a network involving the cuneus for the dual-task condition. Conclusion: Our results suggest that brain networks associated with gait control vary according to walking speed and depend on each walking condition. Gait control in aging involved a distributed network including regions for emotional control that are recruited in challenging walking conditions.