This paper describes an experimental study of two joint configurations, one with only reinforced polymer concrete (RPC) infill, the other with RPC infill and bonded GRP plates, to overcome shear and moment discontinuities in GRP bridge decks transverse to the bridge's main girders. Fabrication, instrumentation and combined moment/shear force loading of these joints are reported. The joints occupied the full 0.9 m widths and the central 1 m lengths of simply supported deck specimens of span 2.7 m. Cutting out of GRP from the deck to create holes for pouring the concrete into the voids of the deck did not appear to compromise joint integrity under short-term load. Rosettes on the webs of the deck lead to the deduction that over 90% of shear force in the joint was carried by the RPC infill at only 30% of the length of the joint inwards from one end of the infill. Beam theory reliably predicts the neutral axis heights in both joint sections after cracking of the polymer concrete, and also predicts the cracking load for the joint with only RPC infill. For the joint with RPC infill and GRP plates, the test data appear to be elusive in showing a clear cracking load. The GRP plates significantly improved cracked section stiffness beyond that for the RPC alone. A peak load of 500 kN over a 300 mm×300 mm patch damaged but did not fail the deck specimen with either joint. For the RPC-only joint, an unexplained 45% drop in strain occurred in the compression concrete during sustained load over 80 minutes. Practical fabrication issues and further potential research into these joints are discussed. © 2010.