A dataset consisting of 70 surface pollen sam- ples from forest, alpine meadow, alpine steppe, temperate steppe, desert steppe, shrub/semi-shrub steppe and desert sites in the Tianshan Mountains, northwestern China pro- vides an opportunity to study the relationships between surface pollen assemblages and modern vegetation and climate in this region. Redundancy analysis (RDA), the human influence index (HII) and pollen ratios were used to facilitate analysis of the pollen data. The modern pollen assemblages are primarily composed of Picea, Artemisia, Chenopodiaceae, Poaceae, Asteraceae, Nitraria and Ephedra. The results suggest that the surface pollen assemblages of different vegetation types largely represent the modern vegetation in terms of the primary taxa and dominant types. The RDA indicates that the mean annual precipitation (MAP) and the July temperature (TJuly) are the major climate variables that control the modern pollen assemblages. Picea, Artemisia, Poaceae, Cyperaceae, Fabaceae, Asteraceae, Polygonaceae and Apiaceae pollen assemblages are positively correlated with MAP and negatively correlated with TJuly, while the pollen ratios for certain other types, such as Chenopodiaceae, Ephedra and Nitraria, are negatively correlated with MAP and positively correlated with TJuly. The arboreal/non-arboreal ratios are notably high in the forest samples, indicating a sensitive response to forest vegetation. Moreover, the Artemisia/Chenopodiaceae pollen ratios are generally cor- related with the vegetation type and annual precipitation change, suggesting that these factors could be useful indicators of moisture variability in arid regions. However, it is difficult to distinguish between steppe and steppe de- sert based on this ratio, due partly to human disturbance. The HII is significantly correlated with certain pollen taxa, including Poaceae, Plantago, Polygonaceae and Elaeag- naceae, particularly in the alpine meadow and steppe samples. Our results have implications for interpreting the available fossil pollen data in the study region and other arid and semi-arid regions.