This study investigated the behavior of bolted shear connections for fiber-reinforced polymer (FRP)–concrete hybrid beams. Fourteen push-out tests were conducted with varying parameters of bolt strength, bolt diameter ($d_{\mathrm{bolt}}$ ), bolt-embedding length in concrete ($h_{\mathrm{bolt}}$ ), FRP shear strength ($S_{FRP}$ ), and the use of ultra-high performance concrete (UHPC). Tests showed that a larger $h_{\mathrm{bolt}}$ leads to smaller slip but has only a slight effect on capacity; specimens with a larger $d_{\mathrm{bolt}}$ have greater capacity and smaller slip. Specimens with ordinary-strength bolts (OSBs) failed at bolt shank shear, whereas specimens with high-strength bolts (HSB) failed at FRP flange shear-out associated with a significant improvement in capacity compared with those with OSBs. The use of UHPC substantially reduced slip, but did not improve bolt capacity because the two failure modes do not occur in concrete. The prediction equations for capacity associated with these two failure modes were derived by modifying the existing equations. A salient finding in the load-slip curves of HSB specimens was three loading stages with different slopes: compacting, hardening, and softening. This was different from observations in previous studies on head studs in steel-concrete composite beams and FRP bolts/dowels in FRP-concrete hybrid beams. The load-slip curves of specimens with HSB were fitted by a trilinear model to facilitate design and analysis. Finally, design recommendations for bolted connections were formulated after parametric analysis.