We constrain the abundance of primordial black holes ( PBH ) using 2622 microlensing events obtained from 5-years observations of stars in the Galactic bulge by the Optical Gravitational Lensing Experiment ( OGLE ) . The majority of microlensing events display a single or at least continuous population that has a peak around the light curve timescale t _ { E } \simeq 20 ~ { } { days } and a wide distribution over the range t _ { E } \simeq [ 1 , 300 ] ~ { } { days } , while the data also indicates a second population of 6 ultrashort-timescale events in t _ { E } \simeq [ 0.1 , 0.3 ] ~ { } { days } , which are advocated to be due to free-floating planets . We confirm that the main population of OGLE events can be well modeled by microlensing due to brown dwarfs , main sequence stars and stellar remnants ( white dwarfs and neutron stars ) in the standard Galactic bulge and disk models for their spatial and velocity distributions . Using the dark matter ( DM ) model for the Milky Way ( MW ) halo relative to the Galactic bulge/disk models , we obtain the tightest upper bound on the PBH abundance in the mass range M _ { PBH } \simeq [ 10 ^ { -6 } , 10 ^ { -3 } ] M _ { \odot } ( Earth-Jupiter mass range ) , if we employ “ null hypothesis ” that the OGLE data does not contain any PBH microlensing event . More interestingly , we also show that Earth-mass PBHs can well reproduce the 6 ultrashort-timescale events , without the need of free-floating planets , if the mass fraction of PBH to DM is at a per cent level , which is consistent with other constraints such as the microlensing search for Andromeda galaxy ( M31 ) and the longer timescale OGLE events . Our result gives a hint of PBH existence , and can be confirmed or falsified by microlensing search for stars in M31 , because M31 is towards the MW halo direction and should therefore contain a much less number of free-floating planets , even if exist , than the direction to the MW center .