In practice, deep beams with web reinforcement are the predominant form of deep beams. Understanding the effect of web reinforcement in deep beams made of lightweight aggregate concrete (LWAC) is essential for the structural application of LWAC deep beams, but it is still a pending subject. In this study, eight LWAC deep beams and two normal weight concrete deep beams were tested to investigate the influence of web reinforcement and beam depth on the shear behavior and size effect of LWAC deep beams. The experimental results showed that all deep beams failed in shear–compression mode, which is independent of the web reinforcement ratio and beam depth. The web reinforcement confined the concrete within the strut area and limited the concrete spalling, which positively affected the normalized serviceability/ultimate shear strength. Moreover, web reinforcement effectively mitigated the size effect by resisting the transverse tensile stress and improving the boundary confinement of the concrete struts, but could not eliminate the size effect in LWAC deep beams. Based on these results, a modified strut-and-tie model (MSTM) was developed to quantify the contribution of web reinforcements to the shear capacity more accurately than the conventional strut-and-tie model. The predictions obtained using the MSTM with the Warwick-Foster’s strut efficiency factor were in excellent agreement with the experimental results.