There is an emerging shift in human–computer interaction (HCI) research from things to events and towards time and temporality as a design material, which is made even more urgent by the unique time of the COVID-19 period. This paper pushes this shift forwards by investigating factors and the way that these shape online media multitasking behaviour over time during COVID-19. We model the factors along the WHAT and HOW dimensions of the HCI-over-Time model (HCIoT) with self-report data from 117 university students and objective behavioural data from 40 university students, who participated in an online course over 2 weeks during COVID-19. The results indicated a pervasiveness of media multitasking behaviour over time in an online course, driven by individual factors and enhanced by their mutual fit. Based on interpretation of our data, we suggest conceptualizing the COVID-19 period as the larger temporal environment in the HCIoT model. The discussion further explains how the broader idea of human–computer-environment fit is significant to understand HCIoT through an interaction lens. We discuss methodological issues related to differentiating between self-report and behavioural measures when applying the HCIoT model. The conclusion supports the feasibility and significance of conceptualizing media multitasking during COVID-19 as temporal HCI and of further developing and operationalizing the HCIoT model by using both behavioural and self-report measures.