A midbody fracture of a proximal sesamoid bone (PSB) is a severe injury, especially in racehorses. Screw fixation may be the only option for horses to return to athletic activity. In a cadaveric biomechanical study, fixation of simulated transverse midbody PSB fractures with a 4.5 mm headless compression screw (HCS) was compared to the fixation using a 4.5 mm cortex screw (CS) inserted in lag fashion. The front limbs of 8 horses were prepared with a standardized midbody transverse osteotomy in each medial PSB. The left or the right limb of each pair was randomly assigned to the CS or HCS group. Fracture reduction and fixation was controlled by radiography. Markers were fixed proximal and distal to the osteotomy to document gap opening by an image correlation measurement. Cyclic compressive loading was applied with a hydraulic cylinder in an axial manner leading to tensile cyclic loading in the PSBs. After mechanical testing all limbs were radiographed to document the mode of failure. The mean cycles to total failure were 27,803 for the HCS group and 36,624 for the CS group, respectively. A mean of 14,444 cycles (HCS group) versus 27,464 cycles (CS group) was recorded at a fracture gap opening of 10 %. The mean value of initial stiffness was 300 N/mm (HCS group) versus 590 N/mm (CS group). On post-testing radiographs, failure by screw breakage was detected in 3/8 limbs of the HCS group and in 2/8 limbs of the CS group. Screw pullout and bone failure was noted in 5 limbs in each group. In one limb of the CS group, the medial suspensory branch was disrupted and the fracture fixation was intact. None of the differences between HCS and CS constructs reached statistical significance. However, the lower absolute numbers of cycles sustained before failure, at 10 % gap opening and lower initial stiffness of HCS constructs as well as limitations of the study warrant further investigation before clinical use of the HCS for this indication can be recommended.