This study investigates how the course format change caused by covid-19 pandemic affected learning behaviors and performance of college students enrolled in a large introductory history course. Clickstream log files capturing how students were interacting with online learning contents were analyzed to identify the learning behaviors of students before and after the mid-semester course format change. The non-parametric regression model was developed to examine the relationship between learning performance of students and course format change. Although the frequency of accessing learning resources decreased during the first three weeks after the course format change, it had a relatively small effect on the learning performance of students. The quantile regression model indicates the mid-semester course format change is associated with about 3.3% decrease in the learning performance of students. These results suggest that students were quite resilient and their learning during the pandemic was not as bad as we feared.