Background: This is an updated version of the original Cochrane review published in The Cochrane Library, Issue 1, 2002.Yoga may induce relaxation and stress reduction, and influence the electroencephalogram and the autonomic nervous system, thereby controlling seizures. Yoga would be an attractive therapeutic option for epilepsy if proved effective. Objectives: To assess whether people with epilepsy treated with yoga:(a) have a greater probability of becoming seizure free;(b) have a significant reduction in the frequency or duration of seizures, or both; and(c) have a better quality of life. Search methods: We searched the Cochrane Epilepsy Group Specialized Register (26 March 2015), the Cochrane Central Register of Controlled Trials (CENTRAL, The Cochrane Library, 26 March 2015), MEDLINE (Ovid, 1946 to 26 March 2015), SCOPUS (1823 to 9 January 2014), ClinicalTrials.gov (26 March 2015), the World Health Organization (WHO) International Clinical Trials Registry Platform ICTRP (26 March 2015), and also registries of the Yoga Biomedical Trust and the Research Council for Complementary Medicine. In addition, we searched the references of all the identified studies. No language restrictions were imposed. Selection criteria: The following study designs were eligible for inclusion: randomised controlled trials (RCT) of treatment of epilepsy with yoga. Eligible participants were adults with uncontrolled epilepsy comparing yoga with no treatment or different behavioural treatments. Data collection and analysis: Three review authors independently selected trials for inclusion and extracted data. The following outcomes were assessed: (a) percentage of people rendered seizure free; (b) seizure frequency and duration; (c) quality of life. Analyses were on an intention-to-treat basis. Odds ratio (OR) with 95% confidence intervals (95% Cl) were estimated for the outcomes. Main results: Two unblinded trials recruited a total of 50 people (18 treated with yoga and 32 to control interventions). Antiepileptic drugs were continued in all the participants. Baseline phase lasted 3 months in both studies and treatment phase from 5 weeks to 6 months in the two trials. Randomisation was by roll of a die in one study and using a computerised randomisation table in the other one but neither study provided details of concealment of allocation and were rated as unclear risk of bias. Overall, the two studies were rated as low risk of bias (all participants were included in the analysis; all expected and pre-expected outcomes were reported; no other sources of bias). The overall OR with 95% confidence interval (CI) was: (i) seizure free for six months - for yoga versus sham yoga ORs of 14.54 (95% CI 0.67 to 316.69) and for yoga versus no treatment group 17.31 (95% CI 0.80 to 373.45); for Acceptance and Commitment Therapy (ACT) versus yoga ORs of 1.00 (95% Cl 0.16 to 6.42; (ii) reduction in seizure frequency - the Mean Difference between yoga versus sham yoga group was -2.10 (95% CI -3.15 to -1.05) and for yoga versus no treatment group -1.10 (95% CI -1.80 to -0.40); (iii) more than 50% reduction in seizure frequency - for yoga versus sham yoga group ORs of 81.00 (95% CI 4.36 to 1504.46) and for the yoga versus no treatment group 158.33 (95% CI 5.78 to 4335.63); ACT versus yoga ORs of 0.78 (95% Cl 0.04 to 14.75); (iv) more than 50% reduction in seizure duration - for yoga versus sham yoga group ORs of 45.00 (95% CI 2.01 to 1006.75) and for yoga versus no treatment group 53.57 (95% CI 2.42 to 1187.26); ACT versus yoga ORs of 0.67 (95% Cl 0.10 to 4.35). In addition in Panjwani 1996 the authors reported that the one-way analysis of variance revealed no statistically significant differences between the three groups. A P-Lambda test taking into account the P values between the three groups also indicated that the duration of epilepsy in the three groups was not comparable. No data were available regarding quality of life. In Lundgren 2008 the authors reported that there was no significant difference between the yoga and ACT groups in seizure free rates, 50% or greater reduction in seizure frequency or seizure duration at one year follow-up. The yoga group showed significant improvement in their quality of life according to the Satisfaction With Life Scale (SWLS) (P < 0.05), while the ACT group had significant improvement in the World Health Organization Quality of Life-BREF (WHOQOL-BREF) scale (P < 0.01). Authors' conclusions: Study of 50 subjects with epilepsy from two trials reveals possible beneficial effect in control of seizures. Results of the overall efficacy analysis show that yoga treatment was better when compared with no intervention or interventions other than yoga (postural exercises mimicking yoga). There was no difference between yoga and Acceptance and Commitment Therapy. However no reliable conclusions can be drawn regarding the efficacy of yoga as a treatment for uncontrolled epilepsy, in view of methodological deficiencies such as limited number of studies, limited number of participants randomised to yoga, lack of blinding and limited data on quality-of-life outcome. Physician blinding would normally be taken to be the person delivering the intervention, whereas we think the 'physician' would in fact be the outcome assessor (who could be blinded), so that would be a reduction in detection bias rather than performance bias. In addition, evidence to inform outcomes is limited and of low quality. Further high-quality research is needed to fully evaluate the efficacy of yoga for refractory epilepsy.