Biomechanically, falling after a walking perturbation may be influenced by:​ (1) the pre-perturbation state of mechanical stability (e.g., stability margins) and (2) the response to a perturbation (i.e., recovery skill). Walking stability margins must be modifiable to serve as a target for fall-prevention interventions. We investigated if neurotypical adults could proactively modulate pre-perturbation anteroposterior stability margins while walking. Eleven adults walked on a treadmill at three speeds with and without anterior and posterior perturbations. We measured stability margins anteriorly at mid-swing and posteriorly at foot strike for pre-perturbation steps. A repeated-measures factorial ANOVA evaluated main effects and interactions of walking speed (0.6, 0.8, 1.0 stats/s) and perturbation type (anterior, none, posterior). With posterior perturbation threats, the posterior stability margins were more positive at foot strike (p < 0.01) compared to trials without perturbations. With anterior perturbation threats, the anterior stability margins were not different at mid-swing compared to trials without perturbations (p > 0.05). With any perturbation threat, step lengths shortened (p < 0.01) and step rates increased (p < 0.01). Step width was not different (p > 0.11). At slow speeds with posterior perturbation threats, double-support time decreased (p = 0.04). Proactive modifications to stability margins are indeed possible in a neurotypical population. Consequently, anteroposterior stability margins may be a feasible target for fall-prevention interventions by targeting decreased step lengths or increased step rates within a given walking speed. We do not know the extent to which the observed effects have a meaningful effect on perturbation recovery.