Currently, gravitational microlensing survey experiments toward the Galactic bulge field utilize two different methods of minimizing blending effect for the accurate determination of the optical depth τ. One is measuring τbased on clump giant (CG) source stars and the other is using `Difference Image Analysis (DIA)' photometry to measure the unblended source flux variation. Despite the expectation that the two estimates should be the same assuming that blending is properly considered, the estimates based on CG stars systematically fall below the DIA results based on all events with source stars down to the detection limit. Prompted by the gap, we investigate the previously unconsidered effect of companion-associated events on $τ$ determination. Although the image of a companion is blended with that of its primary star and thus not resolved, the event associated with the companion can be detected if the companion flux is highly magnified. Therefore, companions work effectively as source stars to microlensing and thus neglect of them in the source star count could result in wrong τestimation. By carrying out simulations based on the assumption that companions follow the same luminosity function of primary stars, we estimate that the contribution of the companion-associated events to the total event rate is ~5f_{bi}% for current surveys and can reach up to ~6f_{bi}% for future surveys monitoring fainter stars, where f_{bi} is the binary frequency. Therefore, we conclude that the companion-associated events comprise a non-negligible fraction of all events. However, their contribution to the optical depth is not large enough to explain the systematic difference between the optical depth estimates based on the two different methods.

4 pages, 1 figure, 1 table, ApJ, submitted