This study empirically evaluates the effectiveness of cold wave adaptation policies, specifically emergency warning messages and designated cold wave shelters, in reducing cold-related illnesses in South Korea. Using monthly panel data from 229 municipalities during the winter seasons (November March) from 2014 to 2023, we applied a two-part model to estimate policy effects separately on (1) the probability of cold-related illness occurrence and (2) the conditional number of patients given illness occurrence. Heterogeneity analyses based on regional aging levels and urban – rural classification were conducted to further examine spatial variation in adaptation effectiveness. The results indicated that warning messages significantly reduced both the probability of illness occurrence and the conditional number of patients. However, their marginal effectiveness diminished as issuance frequency increased, suggesting a nonlinear relationship consistent with information fatigue. In contrast, shelter capacity did not significantly affect the probability of illness occurrence but reduced the number of patients once illness occurred, suggesting that shelters primarily function as post-occurrence mitigation measures rather than preventive tools. Heterogeneity analyses showed that the effectiveness of both policies weakened in highly aged regions and that shelter effects were relatively limited in urban areas. These findings suggest that adaptation effectiveness depends not only on the provision of physical facilities or information but also on how policies are implemented within structurally differentiated regional contexts. In rapidly aging areas, adaptation strategies should therefore incorporate measures to improve accessibility and effective use, including mobility support and targeted communication approaches.