New paleomagnetic data from Permian red beds and Middle Jurassic limestones from the Tarim Basin pose a paradox. Their declinations are similar to Upper Carboniferous to Neogene rocks collected from the same sections, and their inclinations parallel present values. When assuming that lower than expected inclinations in continental sedimentary rocks arise from inclination shallowing effects, then the paleolatitudes of all Upper Carboniferous to Present rocks from Tarim are indistinguishable. Local vertical axis block rotations occurring in the last 20 million years explain why declinations vary at different localities in the basin. Our Middle Jurassic data positions Tarim 23.6 ± 8.4° farther south than that predicted from the coeval reference pole for Eurasia; however, no geologic argument exists to support the closure of a large ocean basin between Tarim and Siberia since the Middle Jurassic. Thus the paradox: are the rocks from Tarim totally overprinted, or is the middle Mesozoic part of the reference Eurasian apparent polar wander path erroneous? Several lines of evidence suggest the Tarim rocks are not remagnetized. We conclude that Tarim has experienced little or no apparent polar wander since the Carboniferous. Moreover, our Middle Mesozoic reconstruction of Eurasia using the new Middle Jurassic pole from Tarim results in a more geologically compatible solution for the eastern Asian blocks over previous reconstructions.