We present a magnetostratigraphic study of the late Mesoproterozoic Malgina and Linok Formations, located along the southeastern (Uchur-Maya region) and northwestern (Turukhansk region) margins of the Siberian craton, respectively. Biostratigraphic, radiometric, and chemostratigraphic data indicate that these formations are likely coeval between 1050 and 1100 Ma. Paleomagnetic analyses reveal a high-temperature component carried by magnetite and/or hematite. This component yields positive fold and reversal tests, together with a positive conglomerate test for the Malgina Formation, which indicates that the magnetization was acquired during or soon after sediment deposition. The mean paleomagnetic direction obtained from the Uchur-Maya region, which is unambiguously representative of the Siberian craton, indicates that it could not have been part of Rodinia at that time if Siberia was located in the Southern Hemisphere and if we assume that Laurentia and Siberia were connected along their present northern shorelines. We emphasize that Siberia could have been part of Rodinia during the late Mesoproterozoic if southern Siberia was joined to the northern part of Laurentia as recently proposed by Rainbird et al. <1998>. If true, placing the Siberian craton in the Southern Hemisphere implies that the magnetic polarity of the ~1000 Ma Laurentian paleomagnetic poles must be switched. Our data also show the occurrence of at least 15 symmetric geomagnetic field reversals, indicating that the paleomagnetic results from the late Mesoproterozoic Keweenawan lavas do not reflect a worldwide and persistent asymmetric field during the Proterozoic.