Description: A systematic bias towards low palaeomagnetic inclination recorded in clastic sediments, that is, inclination shallowing, has been recognized and studied for decades. Identification, understanding and correction of this inclination shallowing are critical for palaeogeographic reconstructions, particularly those used in climate models and to date collisional events in convergent orogenic systems, such as those surrounding the Neotethys. Here we report palaeomagnetic inclinations from the sedimentary Eocene upper Linzizong Group of Southern Tibet that are ~20° lower than conformable underlying volcanic units. At face value, the palaeomagnetic results from these sedimentary rocks suggest the southern margin of Asia was located ~10°N, which is inconsistent with recent reviews of the palaeolatitude of Southern Tibet. We apply two different correction methods to estimate the magnitude of inclination shallowing independently from the volcanics. The mean inclination is corrected from 20.5° to 40.0° within 95 per cent confidence limits between 33.1° and 49.5° by the elongation/inclination ( E / I ) correction method; an anisotropy-based inclination correction method steepens the mean inclination to 41.3 ± 3.3° after a curve fitting- determined particle anisotropy of 1.39 is applied. These corrected inclinations are statistically indistinguishable from the well-determined 40.3 ± 4.5º mean inclination of the underlying volcanic rocks that provides an independent check on the validity of these correction methods. Our results show that inclination shallowing in sedimentary rocks can be corrected. Careful inspection of stratigraphic variations of rock magnetic properties and remanence anisotropy suggests shallowing was caused mainly by a combination of syn- and post-depositional processes such as particle imbrication and sedimentary compaction that vary in importance throughout the section. Palaeolatitudes calculated from palaeomagnetic directions from Eocene sedimentary rocks of the upper Linzizong Group that have corrected for inclination shallowing are consistent with palaeolatitude history of the Lhasa terrane, and suggest that the India–Asia collision began at ~20°N by 45–55 Ma.