<p>Handstands are a key skill in many specialist sports involving the use of upper limbs in a role usually associated with lower limbs. Handstand sports have a high incidence of pain in the shoulder, but with little understanding of handstand walking biomechanics. We aimed to explore the biomechanics of handstand walking in athletes with and without shoulder pain to assess underlying differences across whole limb function. Ten participants, who could handstand walk 6 steps, had force and positional data collected. Shoulder pain was assessed via questionnaire. Forces, centre of mass position, velocities, axial and torsional work, mechanical energy, and overall arm parameters were calculated from 237 stance phases. Two-way ANOVA assessed differences between participants with and without shoulder pain and arm (Right/Left). Participants with shoulder pain moved significantly slower (<i>p</i> = 0.005), exhibited shorter arm lengths (<i>p</i> = 0.013) and had greater net torsional work (<i>p</i> = 0.002) resulting in a lower axial–torsional work ratio. Altered handstand biomechanics in participants with pain demonstrate a more flexed position of the arm, often linked with higher joint loading. Handstand walking mechanics appear to closely resemble groucho-running enabling longer double support reducing limb loading on fatigable upper limbs. Future studies should account for confounders including disciplines, training volume, and gender, providing insight into the implications for upper limb full body loading.</p>