Well-preserved pollen archive in lacustrine sediments has been effectively used as indicators of past vegetation and climate, particularly in alpine lakes as stable "pollen sink" continuously receiving pollen grains from local and regional source ranges. Accurate identification of dispersed pollen grains is essential for interpreting past flora, vegetation, and associated climate changes based on pollen data. In this respect, pollen and spore atlases provide irreplaceable references for conducting pollen analysis in relevant study areas. However, poor documents are available on such atlases in alpine lakes in southeastern Tibetan Plateau of China. Here we present pollen and spore atlas spanning the last deglacial to recent decades in the alpine lakes Jirentso and Yueliang, southwestern China. Pollen and spores were imaged with light microscope (LM) and/or scanning electron microscope (SEM), and were organized following different aperture types and sculpture patterns, facilitating comparison and identification. Totally 66 pollen and spore types (54 genera, 31 families) along with some undermined types are classified. Moreover, some frequently occurring pollen taxa and morphologically similar types have been selected to illustrate identification keys that aid in improving taxonomic level. These include three sections of Quercus; Betula, Carpinus, Corylus, and Alnus; as well as Humulus and Moraceae. Owing to similarities of pollen morphology and sculpture pattern under routine amplification of LM (X400), classifying Quercus into different sections should be cautious, so we recommend to use high resolution SEM to confirm proportion of each section, namely sects. Ilex, Cyclobalanopsis, and Quercus. Based on such proportions, calibrating pollen frequency for Quercus is possible to correct LM-based percentages. We also briefly discuss pollen abundance bias for dominant taxa, such as the over-representation of Pinus and Quercus pollen, which significantly overshadows the relative abundance of local or regional taxa (e. g., Rhododendron), resulting in a poor correlation between pollen percentages and the actual abundance of their source's vegetation. Similar issues of low pollen representation also occur with Larix, Cupressus, and key meadow herbs around the lakes (e. g., Primula, Potentilla, and Ranunculaceae). Furthermore, identifying these poorly preserved pollen types are likely hindered by identification bottlenecks under LM. Therefore, SEM or new techniques (e.g., Fourier transform Raman spectroscopy, Fourier transform infrared spectroscopy, Fluorescence lifetime imaging) and alternative proxies (e. g., sedaDNA, phytolith, stomata, and carbon or/and nitrogen isotope) are expected to overcome these shortages in future study. The atlas is open for future updates, given more pollen types are recognized, nevertheless, the atlas in current form lays a robust foundation for Quaternary pollen analysis not only in our study area and surroundings, but also in relevant regions with similar vegetation types, and should provide a useful reference in pollen-based reconstruction of plant diversity, floristic shifts, vegetation dynamics, as well as their responses to climate change in alpine areas of southwestern China.