Electrically assisted two-wheelers (e-bikes) have become an integral component of urban mobility in China. However, the sensitivity of e-bike accident risk to short-term meteorological variations at fine temporal scales remains underexplored. This study investigated the association between short-term hourly ambient temperature exposure and e-bike accident risk in Guangzhou, China, with particular attention to lag structure and effect heterogeneity. Using 4,041 accident records collected between 2022 and 2024, we applied a time-stratified case-crossover design combined with distributed lag non-linear models (DLNMs) to estimate temperature-associated risk while adjusting for atmospheric confounders. We found a positive association between ambient temperature and e-bike accident risk. The association was strongest within the first 1−2 hours after exposure (RR ≈ 1.023 per 1 °C increase) and gradually attenuated over the subsequent 12−14 hours. Stratified analyses showed that the association was more evident during the daytime period and in winter. Interaction analyses further suggested that the temperature-associated risk increase was more pronounced under low-to-moderate humidity and non-rainy conditions, whereas it appeared weaker under high-humidity or rainy conditions. These findings provide new evidence on the acute, hourly-scale effects of temperature on e-bike traffic safety and support the development of meteorology-informed early warning strategies for high-exposure rider groups in warming urban environments.