Traditional and virtual reality (VR)-based balance training can enhance neuromuscular responses and postural adjustments. This study aimed to compare the effects of traditional versus VR-based balance training on muscle activation and anticipatory and initial postural adjustments in healthy older women in response to anticipated perturbations. In this study, 45 healthy older women (60–70 years) were randomly assigned to three groups: VR, traditional training, and control. Intervention groups performed 24 sessions training. Muscle activity of the tibialis anterior (TA), medial gastrocnemius (MG), rectus femoris (RF), biceps femoris (BF), external oblique (EO), rectus abdominis (RA), and erector spinae longissimus (ESL) was measured using surface electromyography during initial, anticipatory, and compensatory postural adjustments while crossing an obstacle. Data were analyzed using analysis of covariance (ANCOVA) and Bonferroni post-hoc tests. In the anticipatory and initial postural adjustment phases, the activity of the TA (P = 0.044 and P = 0.023, respectively) and MG (P = 0.049 and P < 0.001, respectively) muscles was significantly lower in the VR group compared to the traditional training group. No significant changes were observed in other proximal lower-limb muscles (RF, BF, EO, RA, ESL) (P > 0.05). VR-based balance training may optimize anticipatory and initial ankle muscle responses and improve postural control in older adults. Future training programs should incorporate targeted exercises to activate proximal lower-limb muscles and simulate real-life motor demands. Combining traditional and VR-based training with targeted exercise design could serve as an effective strategy for balance rehabilitation in older adults and fall risk reduction.