The COVID-19 pandemic has forced many college students in developing countries to engage in online learning for the first time, and the sudden transit has raised concerns regarding students' competencies for, perception of, and attitude towards online learning. To address those concerns, this study measured three essential constructs of online learning (self-regulated learning, perceived presences, and learning motivation) based on a national survey in China (N = 12,826) and employed structural equation modeling to investigate their intertwined relationship. The study results reveal that (1) college students' academic achievement cannot effectively predict their self-regulated learning in an online learning context; (2) self-regulation can be further differentiated into general and task-specific strategies with a varying impact on three types of presences; (3) online learning motivation is best predicted by cognitive presence, followed by social presence and teaching presence; and (4) the path of task-specific self-regulated learning [right arrow] cognitive presence [right arrow] online learning motivation generates the largest positive compound effect. Implications for online teaching and learning practice are also discussed through the stakeholder perspectives of students, teachers, and platform developers.