Foot orthoses are thought to improve pain by potentially modifying internal mechanical forces. To test this, we explored whether foot orthoses can modify patterns of bone marrow lesions (BMLs) in people with midfoot pain. Forty-two people were recruited with midfoot pain and MRI-confirmed midfoot BMLs. Participants were randomised (2:1 ratio) to receive either pre-formed orthoses (n=27) or control cushioning insoles (n=15). Outcomes included foot pain (VAS), pain and functional impairment subscales of the Manchester Foot Pain and Disability Index and BML volume measured at baseline and 12 weeks. In total 108 bones in the midfoot were identified with BMLs (mean 2.5 bones, SD 1.6). In the orthoses group, pain significantly reduced at 6 weeks (mean VAS = -14.8mm, CI -22.3 to -7.3) and 12 weeks (mean VAS =-7.1mm, CI -15.0 to -0.9) compared to the control group at 6 weeks (mean VAS =-7.4mm, CI -19.9 to 5.2) and 12 weeks (mean VAS =2.8mm, CI -9.1 to 14.7). In the orthoses group, functional impairment and pain impairment were significantly reduced at 6 weeks and to a lesser extent at 12 weeks. In the control group, only the functional impairment reduced significantly at 6 weeks. At 12 weeks, BML volume reduced more in the orthoses group (-1544.4mm3 CI -3660.4 to 571.6), compared to the control group (-315.8 mm3, CI -1528.2 to 896.7). The foot orthoses group showed a greater reduction in foot pain and a greater reduction of in the volume of BMLs compared with the control group.