There is effective short-term central stimulant (CS) treatment of attention-deficit/hyperactivity disorder (ADHD) symptoms in adult ADHD, and despite this, many patients discontinue the treatment too early. The present study examines the duration of CS treatment in a Norwegian cohort of adult ADHD to identify predictors for long treatment duration (≥ 3 years). The data were based on the medical records of a sample of 117 of all 119 adult ADHD patients diagnosed over a period of 8 years in a specific catchment area and treated with CS. A logistic regression model was applied to identify possible predictors for long treatment duration. The sample was severely impaired in terms of comorbidity, educational achievement, and employment. The median duration of CS treatment was 33.0 months. Use of extended-release formulations of methylphenidate predicted long treatment duration positively (odds ratio, 4.420 [95% CI, 1.875-10.419]), whereas baseline antisocial personality disorder predicted long treatment duration negatively (odds ratio, 0.210 [0.085-0.518]). This study showed that it is possible to treat severely impaired and highly comorbid adult ADHD patients with CS over years. The finding that extended-release formulations of methylphenidate predict long treatment duration supports previous research, and it is important to show this association in naturalistic samples such as in the present study. Comorbid antisocial personality disorder was a negative predictor of long CS treatment duration. Univariate analyses also indicate that comorbid substance use disorder is related to shorter duration of CS treatment.