Rootstock recommendation is complicated by performance-site interactions. The N C140 Regional Project recently completed a lo-year evaluation of 9 rootstocks in locations across North America. Based on this data, we developed stability analysis models and demonstrated significant rootstock-site interactions for cumulative yield (CY) and trunk cross-sectional-area (CSA). The models require a site index (SI) estimated from mean performance over rootstocks within site. Prediction of rootstock performance in untested sites would be possible with an independent estimate of SI. We tested prediction of SI from mean maximum temperature (T) and total moisture received (M) and divided T and M into 5 phenological periods: Dee-Jan (Dormant), Feb-Apr (Prebloom), May-Jun (fruit Set), July-Sept (fruit Growth), and Oct-Nov (Postharvest). SICSA was not predicted by any T or M variable. SICY was predicted by TSet. TGrow, and MSet, but TSet and MSet were codependent. SICY was best predicted from a linear relationship with TSet.