Background. Little information exists about longitudinal changes in spatiotemporal gait asymmetry during rehabilitation, despite it being a common goal. Objectives. To describe longitudinal changes in spatiotemporal gait asymmetry over rehabilitation and examine relationships with changes in other poststroke impairments. Methods. Retrospective chart reviews were conducted for 71 stroke rehabilitation inpatients. Admission and discharge measures of spatiotemporal symmetry, velocity, motor impairment, mobility and balance were extracted and change scores were calculated. Relationships between changes in spatiotemporal symmetry and other change scores were investigated with Spearman correlations. Individuals were divided into four groups (worse, no change-symmetric, no change-asymmetric, improved) based on (1) symmetry/asymmetry at admission and (2) symmetry change scores >minimal detectable change. Differences in change scores between groups were investigated with analyses of covariance using the admission value as a covariate. Results. At admission, 59% and 49% of individuals were asymmetric in swing time and step length, respectively. Of these individuals, 21% and 14% improved swing symmetry or step symmetry, respectively. In contrast, 30% improved gait velocity, 62% improved functional balance and 73% improved functional mobility. Associations between change in swing symmetry and change in paretic limb weight bearing in standing and change in step symmetry and change in velocity were significant. There were no significant differences in change scores between the symmetry groups. Conclusions. The majority of asymmetric stroke patients did not improve spatiotemporal asymmetry during rehabilitation despite the fact that velocity, balance and functional mobility improved. Future work should investigate other factors associated with improved spatiotemporal symmetry and interventions to specifically improve it.