The dataset includes information on hydrochemical parameters and diatom analysis of 72 lakes of the Southern and Middle Urals. Based on this dataset we developed transfer function for quantitative reconstructions of electric conductivity (EC) ranging from 55 to 3780 μS cm-1. The best electric conductivity inference model (r2boot = 0.77, RMSEPboot = 0.21 log10μS cm-1) suitable for quantitative reconstructions was developed by weighted averaging (WA) with classical deshrinking. The application of the new inference model for Holocene EC reconstructions from lake sediments cores of the Southern Urals showed an accordance of the vegetation shifts reflected in pollen records and diatom-inferred EC changes.