1. The species richness-elevation relationship (SRE) is predominantly hump-shaped along terrestrial gradients, but has been less explored in aquatic environments. Chironomids were used to evaluate the generality of the SRE in mountain lakes, and the role of methodological and biological factors in determining its shape. 2. The shape of 39 chironomid SREs distributed worldwide was identified by consensus between statistical and visual methods. A ‘coefficient of methodological integrity’ (Cin) was developed to combine information on sampling effort and homogeneity, and elevational extent in order to quantify the adherence of each dataset to methodological standards known to influence the SRE. Differences in the shape of the SRE between biogeographical regions, biomes and climatic regions were tested using Fisher’s exact tests. A formal meta-analysis was conducted to quantify the overall strength of the SRE, and its association with geographical extent, sampling technique, biogeography, biomes and climate. 3. The SRE presented multiple forms, with considerable variation between identification methods. The most satisfactory datasets (i.e. lowest Cin values), showed predominance of non-linear (low-plateau and hump-shaped) patterns. The Cin explained ~21% of pattern variation. Neither biogeography, nor biomes or climatic regions accounted for differences in the shape of the SRE. 4. The global predominance of non-linear SRE suggests that chironomid richness generally remains high in lakes at mid-elevations, decreasing sharply towards high elevations. As previously known for terrestrial environments, identification of SRE shape is influenced by analytical method. Whenever possible, tailoring the sampling design to increase methodological integrity will reduce uncertainty in the identification of SRE shape in mountain lakes.