The b-value, a critical parameter characterising seismic activity and crustal stress states, shows notable temporal variations within earthquake sequences before and after major events. Investigating these variations is vital for unravelling the processes of crustal stress accumulation and release, thus contributing to seismic hazard assessment and earthquake forecasting. This study focused on three major earthquake sequences: the 2011 Tohoku Pacific Coast earthquake, the 2016 Kumamoto earthquake (Japan), and the 2015 Chilean western coast offshore earthquake. We utilised authoritative earthquake catalogues (JMA for Japan, USGS for Chile), processed magnitude homogenisation to unify types (primarily MW) by screening inconsistent records, applied the Gutenberg-Richter relation, and estimated b-values via the maximum likelihood method with a two-year sliding window to capture temporal dynamics. Key results revealed pre-seismic b-value decreases: Tohoku from 0.8 to 0.4 in 3 months, Kumamoto from 1.0 to 0.6 in 4 months, and Chile from 1.0 to 0.6 in 1 month. Post-seismically, values rebounded to 0.8 (Tohoku in 3 months), 1.0 (Kumamoto in 9 months), and 0.95 (Chile in 11 months). Japan showed more pronounced recovery (0.4–0.5 increase) than Chile (0.3–0.4), reflecting tectonic differences. These variations indicate that pre-earthquake decreases are associated with stress accumulation, whereas post-seismic increases correspond to stress release and adjustment. A systematic comparison reveals that regional tectonic differences exert an influence on the magnitude and duration of such variations, thereby enhancing the utility of b-value analysis in characterising stress states within global seismic zones.