Natural and anthropogenically induced soil erosion can cause serious loss of the archaeological record. Our work shows the value of multi-scalar geoarchaeological study when excavating and re-excavating rockshelters in a highly dynamic sedimentary environment where erosion is prominent. Here we present our work on Kisese II rockshelter, Tanzania, originally excavated in the 1950s and largely unpublished, that preserves an important Pleistocene-Holocene archaeological record integral to understanding the deep history of the Kondoa Rock-Art World Heritage Center. Unlike rockshelters in quiescent tectonic settings, like much of central Europe or South Africa, Kisese II exists in highly dynamic sedimentary environments associated with the active tectonics of the Great Rift Valley system exacerbated by human-induced environmental and climate change. We report on our 2017 and 2019 exploratory research that includes integrated regional-, landscape-, and site-scale geoarchaeological analyses of past and present sedimentary regimes and micromorphological analyses of the archaeological sediments. Historical records and aerial photographs document extensive changes in vegetation cover and erosional regimes since the 1920s, with drastic changes quantified between 1960 and 2019. Field survey points to an increased erosion rate between 2017 and 2019. To serve future archaeologists, heritage specialists, and local populations we combine our data in a geoarchaeological catena that includes soil, vegetation, fauna, and anthropogenic features on the landscape. At the site, micromorphological coupled with chronological analyses demonstrate the preservation of in situ Pleistocene deposits. Comparison of photographs from the 1956 and 2019 excavations show a maximum sediment loss of 68 cm in 63 years or >10% of >6-m-thick sedimentary deposit. In the studied area of the rockshelter we estimate ~1 cm/yr of erosion, suggesting the ongoing removal of much of the higher archaeological sediments which, based on the coarse stratigraphic controls and chronology of the original Inskeep excavations, would suggest the loss of much of the archaeological record of the last ~4000 years. These multi-scalar data are essential for the construction of appropriate mitigation strategies and further study of the remaining stratigraphy.