Until now, the detailed interpretation of the observed microlensing events has suffered from the fact that the physical parameters of the phenomenon cannot be uniquely determined from the available astronomical measurements, i.e. the photometric lightcurves. The situation will change in the near-future with the availability of long-baseline, sensitive optical interferometers, which should be able to resolve the images of the lensed objects into their components. For this, it will be necessary to achieve a milliarcsecond resolution on sources with typical magnitudes K $ga 12$. Indeed, brighter events have never been observed up to now by micro-lensing surveys. We discuss the possibilities opened by the use of long baseline interferometry in general, and in particular for one such facility, the ESO VLT Interferometer, which will attain the required performance. We discuss the expected accuracy and limiting magnitude of such measurements. On the basis of the database of the events detected by the OGLE experiment, we estimate the number of microlenses that could be available for measurements by the VLTI. We find that at least several tens of events could be observed each year. In conjunction with the photometric data, our ability to measure the angular separation between the microlensed images will enable a direct and unambiguous determination of both their masses and locations.