Space-based astrometric missions, such as Hipparcos and the proposed GAIA
and SIM missions, can yield frequently sampled, accurate position measurements
for a large number of stars with uniform accuracy. Using Monte Carlo simulations,
we assess the probability of identifying planets around nearby stars in such
databases. The simulations are realistic and include effects such as a
non-circular, randomly oriented planetary orbits, parallax and proper motion
for the parent star, a Hipparcos-like scanning law and a finite mission length.
The probability of false detections is also assessed. We give a simple
parametrization of the discovery probability, based on the ratio of the
astrometric signature to the accuracy of individual astrometric measurements.
