A major problem in the interpretation of microlensing events is that the only measured quantity, the Einstein time scale t_E, is a degenerate combination of the three quantities one would like to know, the mass, distance, and speed of the lens. This degeneracy can be partly broken by measuring either a "parallax" or a "proper motion" and completely broken by measuring both. Proper motions can easily be measured for caustic-crossing binary-lens events. Here we examine the possibility (first discussed by Hardy & Walker) that one could also measure a parallax for some of these events by comparing the light curves of the caustic crossing as seen from two observatories on Earth. We derive analytic expressions for the signal-to-noise ratio of the parallax measurement in terms of the characteristics of the source and the geometry of the event. For Galactic halo binary lenses seen toward the LMC, the light curve is delayed from one continent to another by a seemingly minuscule 15 seconds (compared to t_E ~ 40 days). However, this is sufficient to cause a difference in magnification of order 10%. To actually extract complete parallax information (as opposed to merely detecting the effect) requires observations from three non-collinear observatories. Parallaxes cannot be measured for binary lenses in the LMC but they can be measured for Galactic halo binary lenses seen toward M31. Robust measurements are possible for disk binary lenses seen toward the Galactic bulge, but are difficult for bulge binary lenses.

Revised to take account of important work by Hardy & Walker (1995)