In most literature, the residual stress distribution is assumed to be a symmetrically 2nd order polynomial with compressive stresses near the surface, balanced by tensile stresses in the centre. This assumption leads directly to the assertion that the thickness of the compressive layer is 21.1% of the total thickness. The present paper experimentally investigates the accuracy of this well-established rule for thermally tempered glass and develops a simple equation for asymmetric stress distributions. The experimental investigation is based on more than 6000 measurements of commercially tempered (and heat-strengthened) soda lime silica glass and provides statistics for the compressive zone depth. The dependency of the thickness, residual stress state, and the effect of asymmetric stress distributions is investigated. Due to the scattering in the experimental results, an FE-model has been applied to clarify the effect from the different parameters. •A large experimental campaign is investigating the compressive zone depth in tempered glass.•Commonly used assumptions are investigated and deviations are reported.•FE models are used for investigating the theoretical influence of different parameters on the compressive zone depth.