Background: Acute traumatic musculoskeletal injuries often result in persistent pain and disability despite physical recovery. Virtual reality (VR) provides an innovative approach for overcoming treatment barriers and may help address psychological risk factors for persistent pain and disability. However, the neural mechanisms underlying VR, particularly in subacute orthopedic pain, are insufficiently understood. Objective: This pilot study examined the feasibility, acceptability, and neural correlates of an 8-week home-based therapeutic VR intervention (RelieVRx) for subacute (textless2 mo postinjury) musculoskeletal pain, using functional near-infrared spectroscopy (fNIRS) to assess changes in prefrontal cortex (PFC) activation and coactivation. Methods: Ten orthopedic patients (mean age 46.8, SD 11.86 years; 7/10, 70% female) completed the RelieVRx intervention and underwent fNIRS and behavioral assessments before and post treatment. Behavioral measures included pain intensity, pain interference, pain catastrophizing, pain anxiety, mindfulness, coping, and pain self-efficacy. fNIRS recorded PFC hemodynamic responses during movement-evoked pain and VR-based relaxation/distraction tasks. Feasibility and acceptability were assessed qualitatively and quantitatively (valid fNIRS recordings, participant feedback). Analyses evaluated pre-post changes in PFC activation, functional coactivation, and correlations with behavioral measures. Results: fNIRS procedures demonstrated high feasibility (74/80, 93% valid recordings), acceptability, and there were no safety concerns. Significant improvements were observed across all behavioral measures, including reduced pain intensity at rest (mean difference [MD]=−2.50, Ptextless.001, d=2.24), and with activity (MD=–3.40, Ptextless.001, d=1.98), decreased pain interference (MD range −3.90 to −4.90, Ptextless.001, d range 1.32 to 2.30), reduced pain anxiety (MD=–32.70, P=.001, d=2.14) and pain catastrophizing (MD=–16.40, P=.003, d=2.13), and improved mindfulness (MD=+3.29, P=.01, d=0.94), coping (MD=+0.59, P=.01, d=1.01), and self-efficacy (MD=15.40, P=.008, d=1.51). fNIRS showed significant posttreatment increases in medial PFC activation (right medial channel: distraction task t=−4.473, P=.004; relaxation task t=–3.036, P=0.02) and enhanced coactivation between medial and lateral PFC regions (distraction task t=−2.784, P=.03). Increased functional coactivation between the right medial and left lateral PFC was negatively correlated with improved mindfulness (r=−0.716, P=.046) and coping scores (r=−0.709, P=.049). Conclusions: This study provides initial evidence of the feasibility and acceptability of integrating fNIRS neuroimaging into clinical VR interventions. Results indicate that engagement with VR therapy was associated with alterations in neural activity in key PFC regions implicated in pain regulation, correlating with significant improvements in pain and behavioral measures. The neural and behavioral changes highlight the potential of VR as a mechanistically informed, scalable nonpharmacological approach to managing subacute pain following orthopedic injuries. These findings justify larger trials that incorporate multimodal neuroimaging to further understand potential mechanistic processes that may underlie VR-based pain therapies.