A new analysis of paleomagnetic data from the mid-Cretaceous (∼110 Ma) ultramafic complex at Duke Island (southeast Alaska) supports large poleward transport of the Insular superterrane relative to North America consistent with the Baja British Columbia hypothesis. Previous paleomagnetic work has shown that the characteristic remanence of the ultramafic complex predates kilometer-scale deformation of the very well developed cumulate layering but that the layering was not horizontal everywhere before the folding. It is possible, however, to estimate paleohorizontal for the Duke Island ultramafic complex because the postremanence deformation of the intrusion occurred about two well-defined and spatially separate fold axes. In such a case the tectonically rotated paleomagnetic directions should be distributed along small circles centered on each of the two fold axes. The ancient field direction will lie on both small circles and therefore will be identifiable as one of their two intersection points. Interpreted this way, the tectonically rotated remanence of the Duke Island ultramafic complex defines a mid-Cretaceous (i.e., ancient) field direction that is within 2° of the paleomagnetic direction found by assuming the cumulate layering was initially horizontal (despite the paleomagnetic evidence to the contrary) and performing the standard structure correction. The inferred mid-Cretaceous paleolatitude of Duke Island is 21.2° (2350 km) anomalous with respect to cratonic North America. This result is concordant with southerly paleolatitudes determined by many other workers from bedded rocks of terranes farther inboard in the Insular and Intermontane superterranes.