Some discrepancies have been pointed by various authors (Popper 1997, Torres & Ribas 2002) in the past few years between stellar evolution model predictions in the 0.7 - 1.1 Solar mass range and results obtained from binary stars with the most accurately known properties. The study of the eclipsing binary UV Psc, with relative accuracy of mass better than 1% for both components, suggests that fixing the mixing length parameter αMLT to its solar value (1.6), a standard hypothesis assumed in the most of stellar evolutionary models, is not correct. This parameter seems to decrease with stellar mass. To conform this possible trend, we intend to use new data from an international programme of spectroscopic and photometric observations of detached eclipsing binaries, mainly with solar-type components.