We present a simple toy model of the distribution of objects responsible for gravitational microlensing. We use Monte Carlo simulations to demonstrate how difficult it is to determine the parameters of the lens mass distribution on the basis of the observed distribution of event time scales. A robust determination requires $ \sim 100 $ events, or more, even if the geometry of lens distribution, and the lens kinematics are known.

19 pages in postscript (four figures included), submitted to ApJ