Wind pressure data have been collected on the tiled roof of a full-scale test house at Silsoe in the UK. The tiled roof was of conventional UK construction with a batten-space and bitumen-felt underlay beneath the interlocking concrete tiles. Pressures were monitored on the outer surface of selected tiles, at several locations within the batten-space, and beneath the underlay. Data were collected both with and without ventilator tiles installed on the roof. Little information appears to exist on the share of wind load between tiles and underlays which creates uncertainty in the design of both components. The present study has found that for the critical design case of maximum uplifts it would be appropriate to assign 85% of the net roof load to the tiles and 15% to the underlay when an internal pressure coefficient of -0.3 is used, and to assign 60% to the tiles and 50% to the underlay when an internal pressure coefficient of +0.2 is assumed (an element of design conservatism is inherent in the apparent 110% net loading indicated by the latter pair of percentage values). These findings indicate that compared with loads implied by BS 6399-2, UK design loads for underlay are currently conservative by 25% whilst tile loads are unconservative by around 20% in ridge and general regions and by around 45% in edge regions on average over roof slopes of $15^{circ}-60^{circ}$.