Knowledge that certain large-scale environments might be better simulated than others, or might favor a specific model configuration, can be very valuable for operational forecasting. Present study involves a detail investigation of variations in skill of MCS rainfall forecast among events characterized with different magnitudes of large-scale forcing, along with variations in forecast skill among events due to the use of different convective parameterizations (Betts-Miller-Janjic and Kain-Fritsch). For this purpose simulations of twenty warm season MCS events over the Upper Midwest performed using a workstation version of the National Centers for Environmental Prediction (NCEP) Eta model with ten kilometer grid spacing are used. In addition, an impact of three different types of adjustments to initial conditions (cold pool initialization, vertical assimilation of mesoscale surface observations and relative humidity adjustment based on radar echo coverage) on rainfall forecast skill is investigated.