We are using a recently developed Cartesian grid method to simulate the unsteady aerodynamics of a fruit fly wing model in hovering flight. The fruit fly simulations are compared with force data and digital particle image velocimetry (DPIV) measurements obtained from experimental investigation by others. The force history data obtained from the computations show similar dynamics to the experimental data, despite giving only 2/3 the force magnitude. When the computational data are scaled by a factor of 3/2, the data match very well except at the beginning of the stroke where the experimental data show a stronger force peak than the computational data. Comparisons of the computational flow field with DPIV measurements of a cross section at 0.65 wing span showed very good agreement, and only small differences supported the discrepancies in the force data. Further investigations suggested that the rounded edges of the thicker computational wing is the main cause for the differences with the experimental data. Additional computations showed that slower accelerations at stroke reversal changes the force peak dynamics at the beginning of the stroke, which demonstrates the sensitivity of the computed result to small changes in the stroke dynamics. Also, an analysis of the averaged computed lift forces indicates that the computational forces are adequate for sustained hovering flight of the fruit fly.