This paper presents an analysis of lensing of unresolved background st
ars. Previous theoritical calculations of the lensing rates and optica
l depths considered only resolved stars. However, if a faint unresolve
d star is lensed close enough to a resolved star, the event will be se
en by the microlensing experiment and attributed to the brighter star.
The blending biases the duration, making the contribution of the unre
solved stars very significant for short events, This contribution migh
t be confused with lensing by brown dwarfs. The exact rates of these b
lended events are extremely sensitive to the limiting magnitude achiev
ed in the microlensing search. Calculations of the optical depth and r
ates are provided here, and illustrated in the case of the DUO and OGL
E experiments. The additional contribution of unresolved stars is very
significant, It probably explains the high optical depth and rates ob
served towards the Galactic Bulge for turn-off stars. The blended unre
solved event can be identified using either the color shift or the lig
ht curve shape. However, neither of these two methods is apropriate to
identify a large number of blended events towards the Bulge. In some
case with good photometry and small impact parameter an indentificatio
n is possible. This is illustrated by the case of the OGLE 5 event, wh
ich clearly appears as a case of lensing of an unresolved star. The re
cent results obtained by the PLANET collaboration indicate that a high
resolution and dense sampling of the light curve is possible, and wil
l make possible to correct the blending bias, as demonstrated for OGLE
5. This possibility, is certainly better than a statistical estimatio
n of the lensing rates, which are always prone to some uncertainty. Bu
t, at this time, except in the case of the MACHO measurement of the op
tical depth for the clump giants, the analysis of Microlensing events
found in the various microlensing experiments requires the uses of mod
elisations of the contribution of unresolved stars.