We predict bond betas conditioning on various macro-finance variables. We explore differences across long-term government bonds, investment grade corporate bonds, and high yield corporate bonds. We conduct out-of-sample forecasting using the new approach of combining explanatory variables through complete subset regressions (CSR). We consider the robustness of CSR forecasts across the 1-month, 3-month, and 12-month forecasting horizon. The CSR method performs well in predicting bond betas. Keywords: bond betas; complete subset regressions; corporate bonds; government bonds; macro-finance variables; model confidence set. JEL Classifications: C22; C53; C55; G12.