Cordilleran metamorphic core complexes (MCCs) and their associated structures (detachment faults and mylonites) are the products of large-magnitude Cenozoic extension. Paleomagnetic data from the Harquahala, Harcuvar, Buckskin, and Whipple MCCs of western Arizona and southeastern California (~1900 samples; 191 of 278 sites accepted for inclusion in group and grand means) allow us to estimate the magnitude of footwall tilting associated with extension. All MCC crystalline rocks sampled are in the footwalls of associated brittle detachment faults. Rocks sampled include synkinematic, Miocene intrusions (dikes and plutons), Late Cretaceous granites, and Proterozoic host rocks. These rocks (with different levels of success at the site level) yield dual polarity, high unblocking temperature, and high to moderate coercivity magnetizations. Demagnetization behavior reveals that those sites with well-grouped magnetizations and high laboratory unblocking temperatures and/or high to moderate coercivities have characteristic remanent magnetizations carried by either magnetite, with minor hematite, or by hematite, with minor magnetite. Results of baked contact tests show that these footwall rocks possess thermoremanent magnetizations or high-temperature thermochemical remanent magnetizations acquired early in their cooling history, either before or during extension. Our overall interpretation of the paleomagnetic data is that during Cenozoic extension, MCC footwalls tilted (flattened) only a modest amount (<15¿) in the direction of tectonic transport. We also interpret Harcuvar and Whipple MCC paleomagnetic data to indicate tectonic tilting (steepening) of the mylonitic zone to form a back dipping mylonitic front. The Whipple-Bullard-Eagle Eye detachment fault system and associated mylonites are interpreted to have originated with a low-angle dip (10¿ to 25¿ NE). We relate the small amount of footwall flattening in the direction of tectonic transport and the steepening of the back dipping mylonitic front to the buoyant rise of exhumed footwalls according to rolling-hinge/isostatic uplift models. Our interpretation refutes the widespread applicability of models that predict MCCs to represent strongly tilted crustal blocks, originally bounded by high/moderate-angle normal faults. Additionally, a fold test of Harquahala MCC data yielded negative results. This suggests that antiformal footwall arches represent the oldest structures of MCCs that developed prior to, or during mylonitization of footwall rocks (before acquisition of the remanent magnetizations carried by these rocks). ¿ 2001 American Geophysical Union