Cold spells have been associated with specific diseases. However, there is insufficient scientific evidence on the effects of cold spells on out-of-hospital cardiac arrest (OHCA). Data on OHCA cases and on meteorological factors and air pollutants were collected between 2013 and 2020. We adopted a quasi-Poisson generalized additive model with a distributed lag nonlinear model (DLNM) to estimate the effect of cold spells on daily OHCA incidence. Backward attributable risk within the DLNM framework was calculated to quantify the disease burden. We compared the effects and OHCA burden of cold spells using nine definitions. The risks of different cold spells on OHCA increased at higher intensities and longer durations. Based on Akaike's information criterion for the quasi-Poisson regression model and the attributable risk, the optimal cold spell was defined as a period in the cold month when the daily mean temperature was below the 10th percentile of the temperature distribution in the study period for at least 2 days. The single-day effect of the optimal cold spell on OHCA occurred immediately and lasted for approximately 1 week. The maximum single-day effect was 1.052 (95% CI: 1.018-1.087) at lag0, while the maximum cumulative effect was 1.433 (95% CI:1.148-1.788) after a 14-day lag. Men were more susceptible to cold spells. Young and middle-aged people were affected by cold spells similar to the elderly. Cold spells can increase the risk of OHCA with an approximately 1-week lag effect. Health regulators should take more targeted measures to protect susceptible populations during cold weather.