Aims: Prominent anxiety symptoms are related to poor clinical course and outcome in major depressive disorder (MDD). The aim of this randomised, open-label, controlled study is to compare the efficacy and tolerability of mirtazapine in the form of orally disintegrating tablets against paroxetine in treating MDD patients with anxiety symptoms. Methods: A total of 60 MDD patients with a score above 18 on the Hamilton Anxiety Rating Scale (HARS) were randomly assigned to 8 weeks of fixed dosing treatment with mirtazapine (15-30 mg/day) and paroxetine (10-20 mg/day). Efficacy was primarily assessed with the HARS and with the 17-item Hamilton Depression Rating Scale (HDRS) at weeks 1, 2, 4 and 8 after treatment. Tolerability was assessed from adverse events. Results: The generalised estimating equations (GEE) models showed that the rates of improvement in HDRS scores from baseline to week 8 were similar between mirtazapine and paroxetine groups. However, patients with mirtazapine exhibited earlier improvement in HARS scores at weeks 1 and 2. Week-by-week GEE models showed that these significant differences in improvement of HARS scores between the two treatment groups were detectable from the first evaluation after the treatment (week 1) and maintained through week 2. There was no difference in the overall frequency of adverse events experienced between the two treatment groups. The most common adverse event in the mirtazapine group was somnolence (n = 8), whereas that in the paroxetine group was gastrointestinal discomfort (n = 9). Conclusions: Mirtazapine and paroxetine were equally effective and well tolerated for the depressive symptoms in MDD patients with the high level of anxiety symptoms. Mirtazapine was, however, more effective in reducing the anxiety symptoms than paroxetine in the early weeks of treatment, suggesting that mirtazapine may have an earlier-onset action for the anxiety symptoms in MDD patients.