We observe tidal triggering of earthquakes by measuring the correlation between the Earth tide and earthquake occurrence. We used the times, locations, and focal mechanisms of the 9350 globally distributed earthquakes with magnitude 5.5 or larger from the Harvard centroid moment tensor catalog. The tidal stress was theoretically computed by using the Preliminary Reference Earth Model and a recent ocean tide model, NAO.99b. We considered the shear stress on the fault plane and the trace of stress tensor, J1. Defining the tidal phase angle at the occurrence time for each earthquake, we statistically tested the phase selectivity using the Schuster's method. For all the earthquakes, no significant correlation is found between the Earth tide and earthquake occurrence both for the shear stress and for J1. By classifying the data set according to fault types, however, we find a high correlation with the shear stress for reverse fault type. The correlation is particularly clear for shallow and smaller earthquakes of this type. Significant correlation with J1 also appears for larger earthquakes of reverse fault type and for shallow and larger ones of normal fault type. We find no correlation for strike-slip type. For all the cases of high correlation, earthquakes tend to occur when the tidal stress accelerates the fault slip, indicating that high correlation is not coincidental but is physically justified. This result strongly suggests that a small stress change due to the Earth tide encourages earthquake occurrence when the stress in the future focal area is near a critical condition.