Globally, expansive urbanization profoundly alters natural habitats and the associated biota. Monitoring biodiversity in cities can provide essential information for conservation management, but the complexity of urban landscapes poses serious challenges to conventional observational and capture-based surveys. Here we assessed pan-vertebrate biodiversity, including both aquatic and terrestrial taxa, using environmental DNA (eDNA) sampled from 109 water sites across Beijing, China. Using eDNA metabarcoding with a single primer set (Tele02), we detected 126 vertebrate species, including 73 fish, 39 birds, 11 mammals, and 3 reptiles belonging to 91 genera, 46 families, and 22 orders. The probability of detection from eDNA varied substantially among species and was related to their lifestyle, as shown by the greater detectability of fish compared to that of terrestrial and arboreal (birds and mammals) groups, as well as the greater detectability of water birds compared to that of forest birds (Wilcoxon rank-sum test p = 0.007). Furthermore, the eDNA detection probabilities across all vertebrates (Wilcoxon rank-sum test p = 0.009), as well as for birds (p < 0.001), were higher at lentic sites in comparison with lotic sites. Also, the detected biodiversity was positively correlated with lentic waterbody size for fish (Spearman p = 0.012), but not for other groups. Our results demonstrate the capacity of eDNA metabarcoding to efficiently surveil diverse vertebrate communities across an extensive spatial scale in heterogenous urban landscapes. With further methodological development and optimization, the eDNA approach has great potential for non-invasive, efficient, economic, and timely assessments of biodiversity responses to urbanization, thus guiding city ecosystem conservation management. [Display omitted] • Surveying urban biodiversity is difficult using traditional approaches. • We mapped pan-vertebrate biodiversity across Beijing using aquatic eDNA metabarcoding. • We identified 126 aquatic and terrestrial taxa spanning fish, reptiles, birds, and mammals. • Waterbody type and size affect eDNA-based biodiversity detection. • eDNA methods enable large-scale urban biodiversity surveillance. [ABSTRACT FROM AUTHOR]