Objective: The primary aim of this study was to evaluate longitudinal pathways to impairment as outlined in the cognitive-behavioral model of ADHD in a sample of 59 college students diagnosed with ADHD. Method: Serial mediation models were used to test whether underachievement, defined as prior year GPA, would longitudinally predict self-reported impairment at the end of the next school year, through negative self-concept and associated changes in symptoms of anxiety and depression, while controlling for baseline impairment and changes in ADHD symptoms. Results: Findings supported the cognitive-behavioral model of ADHD. The association between prior year GPA and overall impairment at the end of the year was fully mediated through self-concept and symptoms of depression. Conclusion: These results help explain why impairment often persists even when ADHD symptoms remit and suggests that internalizing symptoms may be an important target for intervention in college students with ADHD.