Meta-analysis was used to evaluate 4 clinical trials comparing distal spleno-renal shunt (DSRS) with endoscopic sclerotherapy (EVS) in the prevention of variceal rebleeding: the interval between bleeding and therapy ranges from < 14 days to > 100 days. A questionnaire was sent to each author of the published trials concerning methods, definitions and results of the trials in order to obtain more detailed and up-to-date information. The selected end-points for the meta-analysis were: rebleeding, mortality and chronic encephalopathy. Analysis of the results in the questionnaires was made using the method proposed by Collins. The pooled relative risk (i.e. the combined Odds ratio of each trial as an estimate of overall efficacy) of rebleeding was statistically reduced by DSRS (0.16; 95% confidence interval 0.10-0.27). Despite this, the overall risk of death following DSRS was only marginally decreased (0.78; 95% confidence interval 0.47-1.29); the lack of homogeneity in the results does not permit any significant conclusions on this end-point. However, in non-alcoholic patients, the decrease in risk of death was greater, and this without heterogeneity, following DSRS than EVS (0.59; 95% confidence interval 0.23-1.50). The overall risk of chronic encephalopathy was slightly increased after DSRS (1.86; 95% confidence interval 0.90-3.86). In conclusion, DSRS significantly reduced the risk of rebleeding compared to EVS without increasing the risk of chronic hepatic encephalopathy. However, DSRS did not significantly affect the overall death risk. Only in non-alcoholic disease did it seem to show an advantage over EVS.