Regime-switching models are well-suited to capture non-linearities in interest rates. This paper examines the econometric performance of regime-switching models for interest rate data from the US, Germany, and the UK. Strong evidence supports the presence of regime switches, but univariate models are unlikely to yield consistent estimates. Regime-switching models incorporating international short rate and term spread information forecast better, match sample moments better, and classify regimes better than univariate models. Regimes in interest rates correspond reasonably well with business cycles, at least in the US, explaining why regime-switching models forecast interest rates better than single-regime models. Non-linear interest rate dynamics implied by regime-switching models have important implications for macroeconomic literature. The implied volatility and drift functions resemble those estimated using non-parametric techniques. The paper provides an in-depth analysis of the econometric properties of regime-switching models for interest rates in the US, Germany, and the UK. It shows that single-regime models are resoundingly rejected by the data. Two statistical criteria are used to compare alternative one-regime models of short rates to regime-switching models. The first criterion investigates the fit of the models with the unconditional moments of the data. The second criterion concerns the forecasting power of the different models. A new metric is proposed to compare the performance of different regime-switching models in identifying the regime over the sample. The paper also explores the non-linear dynamics implied by the term structure regime-switching models. It finds that regimes in interest rates correspond reasonably well with business cycles, at least in the US. The regime classification implied by regime-switching models is closely related to economic business cycles and the regime ex-ante probabilities are good short-horizon predictors of the business cycles in the US. The paper concludes that regime-switching models are a useful tool for modeling interest rates and that they have important implications for macroeconomic literature.Regime-switching models are well-suited to capture non-linearities in interest rates. This paper examines the econometric performance of regime-switching models for interest rate data from the US, Germany, and the UK. Strong evidence supports the presence of regime switches, but univariate models are unlikely to yield consistent estimates. Regime-switching models incorporating international short rate and term spread information forecast better, match sample moments better, and classify regimes better than univariate models. Regimes in interest rates correspond reasonably well with business cycles, at least in the US, explaining why regime-switching models forecast interest rates better than single-regime models. Non-linear interest rate dynamics implied by regime-switching models have important implications for macroeconomic literature. The implied volatility and drift functions resemble those estimated using non-parametric techniques. The paper provides an in-depth analysis of the econometric properties of regime-switching models for interest rates in the US, Germany, and the UK. It shows that single-regime models are resoundingly rejected by the data. Two statistical criteria are used to compare alternative one-regime models of short rates to regime-switching models. The first criterion investigates the fit of the models with the unconditional moments of the data. The second criterion concerns the forecasting power of the different models. A new metric is proposed to compare the performance of different regime-switching models in identifying the regime over the sample. The paper also explores the non-linear dynamics implied by the term structure regime-switching models. It finds that regimes in interest rates correspond reasonably well with business cycles, at least in the US. The regime classification implied by regime-switching models is closely related to economic business cycles and the regime ex-ante probabilities are good short-horizon predictors of the business cycles in the US. The paper concludes that regime-switching models are a useful tool for modeling interest rates and that they have important implications for macroeconomic literature.