A characteristic magnetization has been isolated from 18 sites in Middle Ordovician igneous and sedimentary rocks from the Builth inlier, Wales, Two conglomerate tests demonstrate that the characteristic magnetization in two of the volcanic units was acquired early, perhaps during initial emplacement. Six sites from the northern and central parts of the inlier give a mean direction of D=153.8¿,I=69.8¿ k=45.8; α95=10¿) that passes a regional fold test. However, 12 sites from the southern end of the inlier, in and near the Llanelwedd quarry, yield a significantly different direction after correction for local tilt: D=176.4¿,I=44.5¿, (k=42.4; α95=6.7¿). The Llanelwedd quarry data are anomalous with respect to both results from the northern parts of the inlier, and those obtained from approximately coeval rocks from the Shelve inlier. We have searched for an explanation for the discrepancy between the magnetization directions of the two areas and have now ruled out some possibilities, including superimposed magnetization components at one of the two areas. The Llanelwedd quarry lies within a wedge of dominantly andesitic volcanic rocks that appear only locally, and which are overlain by two unconformities that are also not present elsewhere in the inlier. One possible explanation for the anomalous nature of the Llanelwedd quarry magnetization is that the strata were not horizontal at the time of magnetization, either due to early volcano-tectonic tilting or primary dips on the flanks of a volcano. Another possibility is that the Llanelwedd quarry rocks were extruded rapidly enough that the geocentric coaxial dipole field hypothesis cannot be applied. Results from Cenozoic volcanic rocks from the western United States suggest that even highly anomalous fields can be preserved virtually unchanged over a number of separate flow events. We currently regard this to be the most viable explanation for the anomalous nature of the Llanelwedd quarry magnetization. We conclude that the data from the 6 northern sites represent the best estimate of the Middle Ordovician axial dipole field for the inlier. ¿ American Geophysical Union 1992