This study aimed to classify latent profiles of Korean undergraduates' academic emotions in an e-learning environment, and to examine the effects of instructional variables on these profiles as well as the differences in their learning outcomes. A survey was conducted among 777 students who participated in online courses offered by a Korean university. Latent profile analysis revealed four types of emotional profiles: a moderate type (MT); a positive type (PT); a negative type (NT); and an ambivalent type (AT). MT comprised 72.5% of the total number of participants and showed medium levels of both positive emotions (PE) and negative emotions (NE). PT comprised 13.1% of the participants and showed high levels of PE and low levels of NE. NT comprised 10.2% of the participants and showed low levels of PE and high levels of NE. AT comprised 4.2% of the participants and both showed high levels of both PE and NE. Further analysis showed that the quality of instructional content, interaction, the system, and evaluation all proved to be predictors of emotional profiles. Moreover, they indicated differences in perceived achievement and in learner satisfaction. Based on these results, this study provides a discussion and suggestions for further studies.