To properly design a moisture durable building envelope, one needs to anticipate how it will respond to indoor and outdoor environmental loads. A common approach is by undertaking hygrothermal simulations. Given the year-to-year variability of climate data, hygrothermal simulations, to be representative, should be performed over several consecutive years. Global warming increasingly being evident, the uncertainty of the future projected climate needs to be taken into account by considering several possible global warming scenarios. This increases the number of simulations to be performed and if the simulations were to be performed for several consecutive years, this would greatly increase the computing time. One way to reduce simulation time is to use moisture reference years. To select moisture reference years, a method is needed to predict or rank the climate years in terms of their anticipated impact on the moisture response of the building envelope. In this study, several models to predict moisture content of cross-laminated timber of a tallwood building wall assemblies were compared: linear stepwise (backward and forward) regression, regularization methods (i.e., LASSO, Ridge and Elastic Net), and Partial least squares regression (PLSR). These models were developed using climate data of several Canadian cities and global warming scenarios. It is shown that for this particular set of data, in respect to predictive ability, linear regression and regularization techniques perform better than PLSR; as to the ability of the different models to rank climate years, all of the models investigated appear to have similar performance.