Maintaining lateral balance during gait primarily involves adjusting foot placements relative to the center of mass (CoM) and its velocity. This foot placement strategy is often impaired in persons with balance deficits and may be improved with rehabilitation. This study assessed the test–retest reliability of common foot placement control metrics in 29 participants with balance impairments (eight stroke, twelve iSCI, nine other neurological conditions) and 22 controls. Participants completed two 2.5-minute walking sessions on an instrumented treadmill on separate days within two weeks. A linear model was used to estimate lateral foot placements from CoM position and velocity. Foot placement control gains (regression coefficients) were derived directly from this model, while foot placement deviation (RMSE) and adherence (R²) were calculated by comparing estimated to observed foot placements. Test-retest reliability was evaluated using intraclass correlation coefficients (ICCs) and Bland-Altman analyses. In persons with balance impairments, foot placement deviation and adherence showed good reliability at initial contact (ICC 0.82 and 0.81, respectively) and moderate to good reliability at midstance (ICC 0.75 and 0.66, respectively), with smallest limits of agreement (LOA) and coefficients of repeatability (RC) at initial contact. Controls showed poor to moderate reliability across both instances (ICC:​ 0.01–0.67), but with lower LOA and RC. Foot placement control gains exhibited poor to moderate reliability in both groups and instances (ICC:​ 0.31–0.63) with large LOA and RC. Hence, foot placement deviation at initial contact was a reliable measure for assessing group-level foot placement control, but high between-measurement errors limit its use for individual monitoring.