The accuracy of two resistive MHD codes based on very different methods is studied by simulating the same problem of plasmoid formation and ejection in the geomagnetic tail initiated by resistive diffusion, using identical initial and boundary conditions and a 2-D approximation. One of the codes is a modified explicit leapfrog scheme, which has been used extensively in the past for similar problems. The other scheme is an implicit particle-in-cell MHD code (FLIP-MHD). The results from both schemes agree within a few percent in all major characteristics: onset and growth rate of the instability, formation and tailward motion of a plasmoid, scale size of the perturbation and details of the mode structures. FLIP, however, predicts a slightly slower tailward motion of the plasmoid. Both simulations exhibit highly variable earthward flow from the reconnection site with similar averages and similar fluctuation levels, resembling observed bursty bulk flows. Due to differences in the phase of these fluctuations, however, the instantaneous flow may differ in the two simulations. ¿ American Geophysical Union 1996