Objective: To conduct a per-protocol analysis on thigh muscle volume outcomes from the Spinal Cord Injury and Physical Activity (SCIPA) Switch-On Trial. Design: Secondary analysis from an assessor-blind randomized, controlled trial. Setting: Four acute/sub-acute hospitals in Australia and New Zealand. Participants: 24 adults (1 female) within four weeks of motor complete or incomplete spinal cord injury (SCI) Intervention: Functional electrical stimulation–assisted cycling (FESC) or passive cycling (PC) 4x/week for 12 weeks. Outcome Measures: Whole thigh and muscle group volumes calculated from manually segmented MR images. Results: 19/24 participants completed ≥ twelve weeks of the intervention. Five participants experienced hypertrophy (4 FESC; 1 PC) and eight attenuation of atrophy (<20% volume loss) (3 FESC; 5 PC) in thigh muscle volume. Six participants were non-responders, exhibiting atrophy >20% (3 FESC; 3 PC). Mean (SD) change for FESC was −2.3% (25.3%) and PC was −14.0% (12.3%). After controlling for baseline muscle volumes, a strong significant correlation was found between mean weekly exercise frequency and quadriceps and hamstring volumes (r=6.25, P=0.006), regardless of mode. Average watts was highly correlated to quadriceps volumes only (r=5.92, P=0.01), while total number of sessions was strongly correlated with hamstring volumes only (r=5.91, P=0.01). Conclusion: This per-protocol analysis of FESC and PC early after SCI reports a partial response in 42% and a beneficial response in 25% of patients who completed 12 weeks intervention, regardless of mode. Strong correlations show a dose–response according to exercise frequency. Characteristics of non-responders are discussed to inform clinical decision-making.