Background: Numerous studies suggest that Internet-based self-help treatments are effective in treating anxiety disorders. Trials evaluating such interventions differ in their screening procedures and in the amount of clinician contact in the diagnostic assessment phase. The present study evaluates the impact of a pre-treatment diagnostic interview on the outcome of an Internet-based treatment for Social Anxiety Disorder (SAD). Method: One hundred and nine participants seeking treatment for SAD were randomized to either an interview-group (IG, N = 53) or to a non-interview group (NIG, N = 56). All participants took part in the same 10-week cognitive-behavioural unguided self-help programme. Before receiving access to the programme, participants of the IG underwent a structured diagnostic interview. Participants of the NIG started directly with the programme. Results: Participants in both groups showed significant and substantial improvement on social anxiety measures from pre- to post-assessment (dIG = 1.30–1.63; dNIG = 1.00–1.28) and from pre- to 4-month follow-up assessment (dIG = 1.38–1.87; dNIG = 1.10–1.21). Significant between-groups effects in favour of the IG were found on secondary outcome measures of depression and general distress (d = 0.18–0.42). Conclusions: These findings suggest that Internet-based self-help is effective in treating SAD, whether or not a diagnostic interview is involved. However, the pre-treatment interview seems to facilitate change on secondary outcomes such as depression and general distress.