Abstract: Clinicians need the most precise and accurate information looking at treatment options in their everyday clinical practice. They require to know the direction and the magnitude of the treatment effect both in some specific trial and across them. It is commonly believed that non-randomized studies give larger estimates of treatment effect than randomized controlled trials (RCTs). However, a recent NHS HTA report found little evidence of a systematic bias comparing both types of studies in four different fields of the clinical practice. In this study we examine the results from systematic reviews of RCTs and non-randomized controlled studies of antiepileptic drugs. Objectives. * To explore potential significant differences in treatment effect estimates amongst different study designs in AEDs trials * To make comparisons between randomized short-term trials versus follow-on open label phases in AEDs trials * To seek potential both trials-related and patients-related causes of the eventual differences in the treatment effect estimates of AEDs across trials Methods. We will make an exploratory meta-analysis, looking at AED add-on placebo-controlled therapy for patients at any age with drug-resistant partial epilepsies. We will be doing a systematic search of trials by the Cochrane Epilepsy Group's trial register, the Cochrane Controlled Trial Register (Cochrane Library Issue 1, 2000) and Medline; unpublished studies will be requested from pharmaceutical companies and colleagues will be asked about any studies that we have missed. We will not be excluding a priori studies with unclear or poor methodology concerning specifically randomization concealment, blinding assessment and intention-to-treat analysis. Patient-level covariates will be undertaken by attempting an associated individual patient data (IPD) meta-analysis. Our primary outcomes will be efficacy (50 % seizure frequency reduction), global effectiveness (proportion of drop-out and withdrawal patients by any reason), side effects, cognitive effects and quality of life.