Binary and multiple systems constitute one of the main tools for obtaining fundamental stellar parameters, such as masses, radii, effective temperatures and distances. One especially fortunate, and at the same time rare, occurrence is that of double-lined eclipsing binaries with well-detached components. In this special case, it is possible to obtain a full solution of all orbital and stellar parameters, with the exception of the effective temperature of one star, which is normally estimated from spectral type or derived from atmospheric analysis of the spectrum. Long-baseline interferometry at facilities such as the ESO VLTI is beginning to have the capability to measure directly the angular separation and the angular diameter of some selected eclipsing binary systems, and we have proposed such observations with the AMBER instrument. In particular, we aim at deriving directly the effective temperature of at least one of the components in the proposed system, thereby avoiding any assumptions in the global solution through the Wilson–Devinney method. We will also obtain an independent check of the results of this latter method for the distance to the system. This represents the first step towards a global calibration of eclipsing binaries as distance indicators. Our results will also contribute to the effective temperature scale for hot stars. The extension of this approach to a wider sample of eclipsing binaries could provide an independent method to assess the distance to the LMC. The observations will extend accurate empirical calibration to spectral type O9 – B0.