Soil freeze–thaw cycles (FTCs) are widespread across mid- to high-latitude and high-altitude regions, with particularly high frequency in permafrost areas where they can enhance soil carbon release and potentially accelerate climate warming. However, the response of soil carbon release to FTCs and its determinants at the regional scale remain unclear due to the lack of direct experimental evidence. Here, based on topsoil samples along a 1000-km transect across the Tibetan permafrost region, we conducted a freeze–thaw microcosm experiment (5 cycles of 1-day freezing and 3-day thawing) to investigate patterns and predictors of FTC-induced soil carbon dioxide (CO₂) release. The results revealed that FTCs significantly increased soil CO₂ release by 20% ± 2% relative to the thaw-only control across all cycles. This response was primarily driven by soil carbon availability rather than microbial properties. Soils with higher carbon availability, as indicated by initially (prior to FTCs) larger soil organic carbon content and less mineral-protected carbon, as well as greater post-thaw increases in dissolved organic carbon, released more CO₂ under FTCs. Such an effect of carbon availability was further confirmed at the global scale. Although FTCs altered the microbial community composition, notably leading to an increase in the proportion of r-strategists, microbial properties played a minor role in influencing the CO₂ release. These results demonstrate the crucial role of soil carbon availability in affecting responses of CO₂ release to FTCs, highlighting the need for models to explicitly characterize pulsed CO₂ release associated with substrate availability to better predict permafrost carbon-climate feedback.