More than 30 years ago a specialist meeting was held at Joint Research Center Ispra (Italy) from 15 to 17 January 1990 to review the current understanding of chemistry during severe accidents in light water reactors (LWR). Let us consider that, at the end of the 1980s, thermodynamics introduced in the severe accident codes was really poor. Only some equilibrium constants for a few simple reactions between stoichiometric compounds were used as well as some simple correlations giving estimates of solidus and liquidus temperatures. In the same time, the CALPHAD method was developed and was full of promise to approximate the thermodynamic properties of a complex thermochemical system by the way of a critical assessment of experimental data, a definition of a simple physical model and an optimisation procedure to define the values of the model parameters. It was evident that a nuclear thermodynamic database had to be developed with that new technique to obtain quite rapidly prominent progress in the knowledge of thermochemistry in the severe accident research area. Discussions focused on the important chemical phenomena that could occur across the wide range of conditions of a damaged nuclear plant. The most pressing need for improved chemical models is identified with condensed phase mixtures to model the corium progression. This paper reviews more than 30 years of experimental data production in the field of corium thermodynamics. This work has been conducted through multiple international programs (EURATOM, ISTC, OECD) as well as through more specific studies conducted at the national scale. This research has been capitalised in specific databases such as NUCLEA and TAF-ID, databases developed at IRSN and at CEA, respectively, and are now used in degradation models of the severe accident simulation codes. This research is presented in this paper. In the conclusion, we outline the research perspectives that need to be considered in order to address today’s and tomorrow’s issues.