BackgroundThe COVID-19 epidemic provides an environment for frequent media multitasking, which might associate with an increase in depression and anxiety. Since many studies have found that media multitasking negatively affects cognitive capacity, we propose a cognitive perspective to explore how media multitasking may associate with mental health. This study examined the potential mediating role of attention control and negative information attentional bias in the relationship between media multitasking and anxiety and depression.MethodsParticipants (n = 567) were recruited from college students in China. They completed an online survey that included the Media Multitasking Inventory (MMI), Attention Control Scale (ACS), Attention to Positive and Negative Information Scale (APNI), Generalized Anxiety Disorder Scale (GAD-7), and Patient Health Questionnaire (PHQ-9). After exploring the correlations between the measures, serial mediation models were examined.ResultsThe results indicated significant positive correlations between media multitasking and anxiety and depression. Media multitasking, anxiety, and depression were negatively correlated with attention focusing, while positively correlated with negative information attention bias. Media multitasking did not correlate with attention shifting. Mediation modeling demonstrated that attention focusing and negative information attention bias played a serial mediating role in the relationship between media multitasking and anxiety and depression. However, the results did not support the serial mediation model through attention shifting and negative information attention bias.ConclusionMedia multitasking does not directly influence anxiety and depression, while attention focusing and negative information attention bias play serial mediating roles in their relationship. This study highlights the potential cognitive mechanisms between media multitasking and anxiety and depression, providing theoretical support for interventions in individual mental health during the epidemic.