Objective To validate simple criteria that distinguish flat flexible from stable supportive walking shoes by comparing their effects on the knee adduction moment (KAM) in people with medial knee osteoarthritis (OA). Design This was a cross-sectional biomechanical study. We proposed five criteria to differentiate flat flexible from stable supportive shoes, and selected three pairs of shoes representing each class for biomechanical testing. 28 participants aged ≥50 years with symptomatic medial knee OA underwent gait analysis barefoot and wearing each of the six selected shoes, in random order. Differences in the peak KAM, KAM impulse and peak knee flexion moment (KFM) across test conditions were evaluated with a two-way repeated measures analysis of variance (ANOVA). Immediate changes in walking pain between conditions were also compared. Results Increases in KAM from barefoot were lower with each of the three flat flexible shoe styles (peak KAM: 6.1–8.9%; KAM impulse: 2.4–5.1%) compared to their stable supportive counterparts (peak KAM: 11.6–15.1%; KAM impulse 10.5–13.2%). There was a significant main effect for footwear class on peak KAM and KAM impulse, whereby stable supportive shoes increased the KAM significantly more than flat flexible shoes (P < 0.001). There were no differences in the KFM or immediate walking pain between footwear classes. Conclusions Our proposed criteria can be used by researchers and clinicians to select flat flexible shoes for people with medial knee OA to minimise knee loading. Future research should evaluate whether wearing shoes based on these criteria translates to improvements in knee OA symptoms and/or slows structural disease progression.