This study presents a simple frequency-dependent regime-switching vector autoregression (VAR) model, where each regime and its associated parameters in the VAR are characterized by their distinct spectral properties. Empirical applications to several key macroeconomic variables reveal clear frequency-dependent switching dynamics, with each regime exhibiting distinctive features regarding spectral properties, volatility, and impulse responses. We compare this model with a conventional regime-switching model (typically studied in the time domain) and highlight several key differences between the two approaches.