Many concrete bridges related to railways in the U.K. consist of prestressed rectangular concrete beams, post-tensioned together transversely to aid lateral distribution of load; this bridge type has been repeatedly flagged as having insufficient shear capacity. Sixteen tests on small-scale beams, which are scaled-down replica models of the actual bridge beams, are presented. The specimens are tested under a four-point loading system and are both prestressed (PRC) with and without stirrups and non-prestressed (RC) with and without stirrups, to provide full understanding of their shear behaviour. Four further tests are then presented on RC beams strengthened in shear with FRP bars inserted from the soffit into pre-drilled holes and fixed in place using epoxy resin; this method allows strengthening in cases where the webs are inaccessible. Comparisons are made with current code predictions for the strength of all specimens. The results show that unstrengthened RC beams behave mostly as expected and as predicted by codes, while for PRC beams a great variation in shear-carrying capacity following shear cracking is observed for different span-to-depth loading ratios. The proposed FRP strengthening scheme is effective and provides significant improvement to the shear-carrying load capacity.