Diatom assemblages are used widely as indicators of environmental conditions. They have been particularly useful in the assessment of some acute environmental problems such as water acidification, eutrophication and salinisation. High mountain lakes are currently a research focus as sentinels of global change, predominantly of long-distance atmospheric pollution and climate warming. The diatom assemblages in these lakes are extremely rich, and the sediment record provides short and long historical perspectives of the changes. We investigated the relative sensitivity of diatoms to the main environmental gradients found in mountain lakes and evaluated the strength and uncertainties for predicting simultaneously several environmental variables associated with these gradients based on a survey of 83 lakes in the Pyrenees. Variables related to the ionic composition and acid-base balance (calcium*, acid neutralizing capacity*, magnesium, sodium, and sulphates), trophic conditions (total phosphorus* and dissolved organic carbon) and physical factors (water temperature*, irradiance at the lake bottom*, and macrophyte cover) explained independently the variation in the diatomassemblages. However, the assemblage predictive capacity of these variables- tested by developing transfer functions (Weighted Averaging- Partial Least Squares) – was only acceptable for a subset of the variables (*). The spatial autocorrelation of the environmental variables had no influence on the performance of the transfer functions except forwater temperature,which is highly dependent on altitude. Our results indicate that diatom assemblages have great potential for assessments of multiple environmental variables in mountain lakes and, consequently, in applications of global change surveillance.