This paper presents experimental results which demonstrate the important, but to date little-studied influence of surfacing on the response up to failure of orthotropic GFRP bridge decking local to concentrated loading. Four bonded deck specimens were tested, two without surfacing and two including a 30 mm thick polymer concrete surfacing layer. A specified steel plate-rubber pad ensemble which produced a load distribution generating high shear and low moment locally was used to load one surfaced and one unsurfaced specimen to failure, while a plate-pad system inducing modest shear and high moment locally (nearer the tyre effect) was used to fail the other two specimens. The surfaced specimens' load capacities exceeded those of their unsurfaced counterparts by 90% and 261% under the shear-dominant and moment-dominant loads respectively. For each surfaced specimen, failure occurred in part by formation of a distinctly curved crack in the polymer concrete. In contrast to reports of detached surfacing on real FRP deck bridges, no debonding was observed at the surfacing-deck interfaces during these tests. This suggests scope for significantly improved surfacing-deck bond integrity in practice. GFRP delamination, particularly within the directly loaded flange, also occurred in the four test specimens. The critical failure zone shifted from the flange's ends to its local midspan in changing from shear-dominant to moment-dominant loading. It is concluded that the status quo standard of testing unsurfaced deck specimens should be modified to testing of surfacing-deck systems. (c) 2012 Elsevier Ltd. All rights reserved.