The Tianshan Mountains have both sunny and shady slopes where the vegetation is quite different and possibly responds differently to climate changes. Lake Xiaolongchi, a small alpine lake located in Northern Tianshan Mountains of arid central Asia, is used to explore the source area of the main pollen taxa in lake sediments. Pollen data of a well-dated sediment core from Lake Xiaolognchi are analyzed to discuss vegetation dynamics on the sunny and shady slopes and their response to climate. The pollen data are used to reconstruct the temperature and humidity changes over the last 8000 years. Picea pollen mainly originates from the shaded slopes, while Artemisia, Amaranthaceae and Poaceae are mainly from the sunny slopes. Cyperaceae pollen originates primarily from the tundra vegetation in the high elevation of the catchment. Accordingly, the taiga biome score reflects changes in arboreal plants on the shady slopes of the catchment, which are sensitive to temperature. High or low taiga biome scores can be used to indicate high or low temperatures, respectively. The steppe and desert biome scores reflect steppe and desert vegetation on the sunny slopes, which are sensitive to humidity. High steppe/ desert biome scores can be used to infer humid/arid climate respectively. The temperature reconstruction based on the taiga biome score exhibits a cooling trend, and the humidity reconstruction based on the steppe and desert biome scores shows a wetting trend over the past 8000 years. These trends are generally consistent with other climate records from the arid central Asia. The climate was extremely dry from 2.1 to 1.7 cal kyr BP, corresponding to the Roman Warm Period (RWP). This extremely dry event appears to be a regional signal and possibly have caused a significant decrease of archaeological sites in arid central Asia in that period.