The objective of this study was to assess the effects of subcutaneous (sc) interferon beta-1a (IFNbeta-1a) on cognition in mildly disabled patients with relapsing-remitting multiple sclerosis (RRMS). Patients aged 18-50 years with RRMS (McDonald criteria; Expanded Disability Status Scale score <or=4.0) were assigned IFNbeta therapy at the physician's discretion and underwent standardized magnetic resonance imaging, neurological examination and neuropsychological testing at the baseline and regular intervals for up to three years. This analysis included 459 patients who received sc IFNbeta-1a (44 mcg: n = 236; 22 mcg: n = 223; three-year follow up was available for 318 patients). The hazard ratio for cognitive impairment over three years (44 mcg versus 22 mcg) was 0.68 (95% confidence interval [CI]: 0.480-0.972), suggesting a 32% lower risk with the higher dose treatment. At year 3, the proportion of patients who were cognitively impaired increased slightly from 23.5% at the baseline to 24.8% in the IFNbeta-1a 22 mcg treatment group, but remained stable at 15.2% in the IFNbeta-1a 44 mcg treatment group. The proportion of patients with cognitive impairment at year 3 was significantly higher in the 22 mcg group than in the 44 mcg group (P = 0.03), although a trend was also seen at the baseline (P = 0.058). Multivariate logistic regression (corrected for baseline cognitive deficits) indicated that treatment with the higher dose of IFNbeta-1a was predictive of lower cognitive impairment at three years (odds ratio: 0.51, 95% CI: 0.26-0.99) compared with the lower dose of IFNbeta-1a. These findings suggest that sc IFNbeta-1a may have dose-dependent cognitive benefits in mildly disabled patients with RRMS, and may support early initiation of high-dose IFNbeta-1a treatment.