This study reexamines the time-varying structure of exchange rate pass-through (ERPT) by employing the dynamic model averaging (DMA) method. DMA is a novel empirical methodology that allows both parameters and predictors of the model to change over time, thereby addressing model uncertainty in determining time-varying ERPT degrees. We apply DMA to quarterly time series data spanning from 1996:1–2021:3 for each of the 39 advanced and emerging economies in the sample. The findings reveal a shift in short run ERPT dynamics prompted by the global crisis in 2008-09. While ERPT levels are initially low, they have been rising steadily since the onset of the pandemic across all advanced countries. Conversely, the ERPT levels are approximately twice as high in emerging economies, exhibiting a U-shaped pattern over the sample period.