AbstractThis paper continues Part I (Wilks DS. Use of stochastic weather generators for precipitation downscaling. WIRES Clim Change 2010, 1(6):898–907) of a two‐part review on statistical downscaling of climate changes using parametric ‘weather generators’, which treated only precipitation downscaling at individual locations. Here the review is extended to include also downscaling of nonprecipitation variables at individual locations, and spatially coherent precipitation and nonprecipitation downscaling. Parametric weather generators are explicitly stochastic models that usually operate on the daily timescale. The use of stochastic methods for climate downscaling is natural and logically consistent because of the inherent indeterminacy of the problem: any number of small‐scale weather sequences may be associated with a given set of larger‐scale values. Downscaled climate changes are simulated by adjusting or varying the parameters of the weather generators, in a manner consistent with dynamically simulated or otherwise assumed larger‐scale climate changes. Two main approaches for such parametric adjustments have been developed, namely changes in the daily weather generator parameters based on imposed or assumed changes in the corresponding monthly statistics, and day‐by‐day changes to the generator parameters that are controlled by daily variations in simulated atmospheric circulation. These methods are reviewed here, and perspectives on their relative merits are offered. WIREs Clim Change 2012 DOI: 10.1002/wcc.167This article is categorized under: Assessing Impacts of Climate Change > Representing Uncertainty