We present deep infrared ($2.2 mu$m) imaging of the Galactic microquasars 1E1740-2942 and GRS 1758-258 using the Keck-I 10-meter telescope in June 1998. The observations were taken under excellent seeing conditions ($sim 0.45 arcsec$ full-width half-maximum), making them exceptionally deep for these crowded fields. We used the USNO-A2.0 catalog to astrometrically calibrate the infrared images (along with an optical CCD image in the case of GRS 1758-258), providing independent frame ties to the known radio positions of the objects. For 1E1740-2942, we confirm potential candidates for the microquasar previously identified by Marti et al., and show that none of the objects near the microquasar have varied significantly from 1998 to 1999. For GRS 1758-258, our astrometry indicates a position shifted from previous reports of candidates for the microquasar. We find no candidates inside our 90% confidence radius to a $2 sigma$ limiting magnitude of $K_s = 20.3$ mag. We discuss the implications of these results for the nature of the microquasar binary systems.