This quasi-natural experimental study examined an online teaching intervention implemented in response to COVID-19 in China in 2020. It applied the difference-in-difference model to examine the impact and path of the intervention on students’ learning performance of a college foreign language (LPCFL). Based on data from records of withdrawing and changing courses, classroom learning, and teaching evaluations; a questionnaire survey of teachers and students; and relevant school documents during the last seven terms, the results indicated that the online teaching intervention could significantly improve students’ LPCFL. This finding remained robust after adopting a placebo test approach to mitigate possible endogeneity issues. Additionally, this study also conducted a group test through sub-sample regression based on students’ discipline characteristics and intervention organization methods. The results showed that the students who participated in the intervention significantly improved in the three disciplines: humanities was most significantly affected, science and engineering were least significantly affected, and economics and management were in the middle. A range effect was observed for organizational methods. The two downward transmission methods by college teaching management terms had significant positive effects, whereas the other two methods of downward transmission by college student management had significant negative effects. An analysis of the action mechanism indicated that the online teaching intervention mostly improved LPCFL through two channels: students’ learning input and learning support. Overall, these findings not only help expand the research framework on macro environmental intervention policy and micro-learning behavior but also have implications for the in-depth understanding of the real learning effect of online learning interventions for college students and their design in the post-COVID-19 era.