To what extent is water quality related to lake size? This was investigated through the analysis of a database concerning 250 French lakes of various types and sizes. The ecological evaluation of these lakes according to the European Water Framework Directive implied the analysis of at least four water samples during one year to determine physico-chemical parameters and phytoplankton composition. The samples were integrated over the euphotic zone and taken at a single station (where depth is maximal). Then, statistically, the question can be the following: is it possible to predict lake size from the analysis of four annual water samples? Random forest models were derived to predict lake volume, mean depth, and surface area from several hundreds of parameters. Mean depth was the most predictable lake dimension (75 % of variance explained) whereas surface area can hardly be inferred from water quality (only 12 %). The variables most related to lake depth were suspended particulate matter concentration, some elemental ratios, the diversity of phytoplankton communities and the abundance of some particular species. Moreover, simulations with the simple mechanistic model LakeMab where carried out to show how phosphorus concentration and its annual variations are influenced by lake depth and area.