This study focuses on the Khyaraany sand/loess/soil section (50.23N 106.73E) in the Northern Mongolian Plateau with the aim of deciphering the paleoenvironmental records and inferring the last glacial Gobi dynamics. The main "ndings are as follows: (1) silt percentage (with a negligible clay percentage) is positively correlated with the organic matter content. (2) The silt (%) and frequency-dependent magnetic susceptibility (%) are negatively correlated. (3) The magnetic susceptibility is positively related to sand percentage and negatively to silt percentage. Based on C dates and the extrapolated ages, the following observations can be made. (1) Unlike the Marine Isotope Stage (MIS) 3 paleosols (24,500, 28,900, 30,700, 34,400 yr BP) that formed under oxidizing-dominated conditions, two MIS 2 paleosols (15,090, 13,030 yr BP) and two Holocene paleosols (8300, 4070 yr BP) were formed under dominantly reducing conditions. (2) Less windy and/or better vegetation conditions generally dominated the later part of the last glacial (15,090-8300 yr BP), during which three paleosols (incipient histosols) were formed (15,090, 13,030, 8300 yr BP). (3) Windy and/or poor-vegetation conditions generally dominated the early part of the last glacial (&24,000-16,000 yr BP). (4) This section documents approximately thousand-year-long redox cycles. Two tentative conclusions can be drawn from this and previous (Feng et al., 1998) studies. (1) The northern boundary of the Gobi has shrunk as many as nine times during the past 40,000 yr: around 34,400, 30,700, 28,900, 24,500, 15,090, 13,030, 8300, 4070 yrBP and the recent 2000+yr. (2) Based on the negative correlation of the susceptibility with silt percentage and organic matter content, it is proposed that the reducing conditions of incipient histosol formations have contributed to the alteration of magnetic minerals from strong forms of oxidized iron to weak forms. Therefore, the magnetic susceptibility is basically an indicator of redox cycles.