In this study, we evaluate eight autoconversion parameterizations against integration of the Kinetic Collection Equation (KCE) for cloud size distributions measured during the NASA CRYSTAL-FACE and CSTRIPE campaigns. KCE calculations are done using both the observed data and fits of these data to a gamma distribution function; it is found that the fitted distributions provide a good approximation for calculations of total coalescence but not for autoconversion because of fitting errors near the drop-drizzle separation size. Parameterizations that explicitly compute autoconversion tend to be in better agreement with KCE but are subject to substantial uncertainty, about an order of magnitude in autoconversion rate. Including turbulence effects on droplet collection increases autoconversion by a factor of 1.82 and 1.24 for CRYSTAL-FACE and CSTRIPE clouds, respectively; this enhancement never exceeds a factor of 3, even under the most aggressive collection conditions. Shining the droplet-drizzle separation radius from 20 to 25 μm results in about a twofold uncertainty in autoconversion rate. The polynomial approximation to the gravitation.collection kernel used to develop parameterizations provides computation of autoconversion that agree to within 30%. Collectively, these uncertainties have an important impact on autoconversion but are all within the factor of 10 uncertainty of autoconversion parameterizations. Incorporating KCE calculations in GCM simulations of aerosol-cloud interactions studies is computationally feasible by using precalculated collection kernel tables and can quantify the autoconversion uncertainty associated with application of parameterizations. Copyright 2009 by the American Geophysical Union.