A sensitivity analysis is performed to test these assumptions. The PSU/NCAR mesoscale model MM5 is used to evaluate the effects of the initial atmospheric moisture availability on storm dynamics and rainfall accumulation for the Northeastern Illinois storm of July 17-18 1996. The atmospheric moisture availability is adjusted over a wide range but within the upper limits of the maximum observed precipitable water using three different moisture adjustment methods. The numerical experiment demonstrates that the relationship of precipitation to precipitable water depends on spatial scale. For large spatial scales, precipitation tends to scale linearly with precipitable water, but with a slope larger than that assumed in PMP analysis. Increasing atmospheric moisture availability increases large-scale wind convergence, which leads to greater average precipitation over the region. Hence, the assumption of constant wind convergence is not valid. The maximized precipitation depends on both the precipitable water ratio and the wind convergence ratio. For small spatial scales, the relationships of precipitation to precipitable water are nonlinear and vary with atmospheric moisture adjustment methods. The maximum precipitation is more associated with the Convective Available Potential Energy (CAPE) than with precipitable water. Unlike at large scales, the simple conceptual model of PMP moisture maximization is inadequate at small scales because the approximate solution of the water budget equation does not hold.