Late Pleistocene records of loess deposition are a critical archive for understanding terrestrial paleo-environment changes in Central Asia. The age of loess is not well known for the deserts regions and surrounding high plateaus in Central Asia. Previous studies have shown that there remains a disparity between ages for loess deposition by luminescence and C-14 dating. This study evaluates the potential of optically stimulated luminescence (OSL) to date a loess sequence resting on fluvial sands in the east Ili Basin, Central Asia. The single-aliquot regenerative-dose (SAR) protocol on coarse grain quartz was employed for equivalent dose determinations. The basal fluvial sand returned a secure OSL age, with low overdispersion value in equivalent doses (19 +/- 2%) of ca. 36 ka and provides a close, but maximum age estimate (within 5 ka) on the initiation of loess deposition. However, the loess yielded high overdispersion values for equivalent doses and age reversals, coincident with diffuse paleosols; indicating that pedoturbation with loess deposition may be a dominant process. OSL ages between ca. 45 and 14 ka calculated using a maximum age model and OSL ages from other sites in the Basin suggests that the latest major period of loess deposition was between 70 and 10 ka ago. A future hypothesis to test based on these analyses is that there may be three periods of heightened loess deposition at ca. 45, 35 to 19 and 14 ka, when desert source areas to the west were particularly dry.