Severe intraventricular hemorrhage caused by extension from subarachnoid hemorrhage or intracerebral hemorrhage leads to hydrocephalus and often to poor outcome. We conducted a systematic review to compare conservative treatment, extraventricular drainage, and extraventricular drainage combined with fibrinolysis. We carried out a search in Medline of the literature between January 1966 and December 1998 and an additional hand-search from January 1990 to December 1998. Pharmaceutical companies were contacted to gather unpublished data. We reviewed the reference lists of all relevant articles. Two authors independently assessed eligibility of the studies and extracted data on characteristics of study design, patients, and treatment. Patients with primary intraventricular hemorrhage were excluded. Main outcome measures were death and poor outcome (defined as death or dependency) at the end of follow-up. No randomized clinical trial has yet been conducted so far, and we therefore reviewed only observational studies. The case fatality rate for conservative treatment (ten studies) was 78%. For extraventricular drainage (seven studies) it was 58% [relative risk versus conservative treatment (RR) 0.74; 95% confidence interval (CI) 0.55-0.99]. For extraventricular drainage with fibrinolytic agents (five studies) the case fatality rate was 6% (RR 0.08; 95% CI 0.02-0.24). The poor outcome rate for conservative treatment was 90%, that for extraventricular drainage 89% (RR 0.98; 95% CI 0.75-1.30) and that for extraventricular drainage with fibrinolytic agents 34% (RR 0.38; 95% CI 0.21-0.68). All RR values remained essentially the same after adjusting for age, sex, World Federation of Neurological Surgeons scale, study design, and year of publication for the studies that provided these data. Outcome is thus poor in patients with intraventricular extension of subarachnoid or intracerebral hemorrhage. This meta-analysis suggests that treatment with ventricular drainage combined with fibrinolytics may improve outcome for such patients, although this impression is derived only from an indirect comparison between observational studies. A randomized clinical trial is warranted.