Earth's climate over the last one million years experienced several ∼100-kyr glacial cycles, but no simple forcing mechanism has been identified. Numerous studies have tried to explain strong 100-kyr glacial cycles without recognizable forcing, which has come to be known as the 100-kyr problem. Few studies have examined 100-kyr band paleoclimatic signals before 1 Ma. A recent study has demonstrated that benthic oxygen and carbon isotope records are phase-locked and amplitude-coupled at the 100-kyr band, but that neither is phase-locked to and amplitude-coupled with the 100-kyr eccentricity signal between 3 and 1 Ma. This phasing and amplitude mismatch of the 100-kyr band between 3 and 1 Ma between marine records and the eccentricity forcing signal has been called the “late Pliocene-early Pleistocene 100-kyr problem.” However, it remains unknown whether terrestrial paleoclimate records are consistent with marine records at the 100-kyr band. Here I show that loess monsoon records from China are amplitude-coupled with benthic oxygen and carbon isotope records at the 100-kyr band, but not with the 100-kyr eccentricity forcing between 3 and 1 Ma. This observation provides further evidence in support of a free 100-kyr oscillation as the cause of the 100-kyr band amplitude variability in paleoclimatic records between 3 and 1 Ma. In contrast, benthic oxygen isotope records and loess monsoon records at the 100-kyr band are not amplitude-coupled with 100-kyr benthic carbon isotope records over the last 0.4 million years, indicating that the late Pleistocene 100-kyr climatic cycles may not result exclusively from a free oscillation.