BACKGROUND Explanations for the effectiveness of stroke units compared with general wards in reducing mortality, institutionalisation and dependence of people with stroke remain undetermined, and the discussion on the most effective stroke unit model is still up for debate. The intensity of non-invasive mechanical monitoring in many western countries is one of the main issues regarding the different models. This is because of its strong impact on the organisation of the stroke unit in terms of the number of personnel, their expertise, the infrastructure and costs. OBJECTIVE To assess whether continuous intensive monitoring compared with intermittent monitoring of physiological variables in people with acute stroke can change their prognosis in terms of mortality or disability. METHODS We searched the Cochrane Stroke Group Trials Register (November 2012), the Cochrane Central Register of Controlled Trials (CENTRAL) (The Cochrane Library 2011, Issue 8), MEDLINE (1966 to November 2012), EMBASE (1980 to November 2012), CINAHL (1982 to November 2012) and the British Nursing Index (1985 to November 2012). In an effort to identify further published, unpublished and ongoing trials we searched trials registers (November 2012) and reference lists, handsearched conference proceedings and contacted trial authors. METHODS We included all randomised, cluster randomised and quasi-randomised controlled trials comparing continuous monitoring with intermittent monitoring in people within three days of stroke onset. We excluded studies confounded by the delivery of care in different settings (that is studies in which the location of the intervention was not in the same ward in the two arms of the trial). METHODS Three review authors independently selected studies for inclusion, assessed methodological quality and extracted data. We sought original data from trialists in two trials and verified the inclusion criteria in another four trials (three presented at conferences and one was from the Chinese Clinical Trial Registry). Where possible, we extracted data on the threshold level of abnormality that triggered intervention for a given physiological variable, the specific intervention given to correct the abnormality and compliance with the allocated therapy. RESULTS Three studies, involving a total of 354 participants, met our inclusion criteria for the primary outcome. Compared with intermittent monitoring, continuous monitoring significantly reduced death and disability at three months or discharge (odds ratio (OR) 0.27, 95% confidence interval (CI) 0.13 to 0.56) and was associated with a non-significant reduction in deaths from any cause at discharge (OR 0.72, 95% CI 0.28 to 1.85). These significant results depend on one study that has a high risk of bias.Continuous monitoring was associated with a non-significant reduction of dependency (OR 0.79, 95% CI 0.30 to 2.06), death from vascular causes (OR 0.48, 95% CI 0.10 to 2.39), neurological complications (OR 0.81, 95% CI 0.46 to 1.43), length of stay (mean difference (MD) -5.24, 95% CI -10.51 to 0.03) and institutionalisation (OR 0.83, 95% CI 0.04 to 15.72) (secondary outcomes). For the last two outcomes we detected consistent heterogeneity across trials.Cardiac complications (OR 8.65, 95% CI 2.52 to 29.66), fever (OR 2.17, 95% CI 1.22 to 3.84) and hypotension (OR 4.32, 95% CI 1.68 to 14.38) were detected significantly more often in participants who received continuous monitoring (surrogate outcomes).We detected no significant increase in adverse events due to immobility (pneumonia, other infections or deep vein thrombosis) in participants who were continuously monitored compared with those allocated to intermittent monitoring. CONCLUSIONS Continuous monitoring of physiological variables for the first two to three days may improve outcomes and prevent complications. Attention to the changes in physiological variables is a key feature of a stroke unit, and can most likely be aided by continuous monitoring without complications related to immobility or to treatments triggered by the relief of abnormal physiological variables. Well-designed, high-quality studies are needed because many questions remain open and deserve further research. These include when to start continuous monitoring, when to interrupt it, which people should be given priority, and which treatments are most appropriate after the identification of abnormalities in physiological variables.