This study aimed to examine freshmen and senior college students' episodic and semantic memory use in classroom context regarding short and long time delays and college experience level. Data were collected in 2014 and 2017, right after students' final exams (T1) and 5 weeks later (T2). Students were given exemplar questions from their final exams and asked whether they remembered a specific learning episode (episodic memory), if they knew the information (semantic memory), or they guessed the answer while answering exam questions. Senior students in 2017 were asked the same set of questions that they had answered in 2014 as T3. The analyses revealed that the ratio of remember responses to all types of responses decreased within 3 years, whereas know ratio remained stable. Moreover, remember-to-know shift occurred only for senior students. This study is important for demonstrating endurance of semantic memory longitudinally, and salience of college experience level cohort sequentially.