Nudging model prognostic fields to observed states derived from re-analyses is a potentially useful way to evaluate model physics parameterizations. This approach is simple to implement and opens the possibility of direct case-study comparisons with independent data from satellites or field campaigns. We will evaluate moist physics variables such as cloudiness and precipitation from CAM5 runs using a variety of nudging strategies. Small changes in vertical profiles of moisture and temperature nudging are seen to make large differences in moist physics variables. These differences can be controlled by making bias adjustments to the target profiles of T and q. A potentially more serious drawback to nudging studies is that the standard implementation of nudging uses linear relaxation terms that act as damping in the model's prognostic equations. Strong nudging dissipates small scale motion and can suppress divergent modes including those involved in intense precipitation. We will discuss an approach that uses time and space filtered forcing derived from standard nudging runs. This forcing can maintain model fields reasonably close to reanalysis values but does not act as damping.