Unilateral spatial neglect causes difficulty attending to one side of space. Various rehabilitation interventions have been used but evidence of their benefit is lacking. To assess whether cognitive rehabilitation improves functional independence, neglect (as measured using standardised assessments), destination on discharge, falls, balance, depression/anxiety and quality of life in stroke patients with neglect measured immediately post-intervention and at longer-term follow-up; and to determine which types of interventions are effective and whether cognitive rehabilitation is more effective than standard care or an attention control. We searched the Cochrane Stroke Group Trials Register (last searched June 2012), MEDLINE (1966 to June 2011), EMBASE (1980 to June 2011), CINAHL (1983 to June 2011), PsycINFO (1974 to June 2011), UK National Research Register (June 2011). We handsearched relevant journals (up to 1998), screened reference lists, and tracked citations using SCISEARCH. We included randomised controlled trials (RCTs) of cognitive rehabilitation specifically aimed at spatial neglect. We excluded studies of general stroke rehabilitation and studies with mixed participant groups, unless more than 75% of their sample were stroke patients or separate stroke data were available. Two review authors independently selected studies, extracted data, and assessed study quality. For subgroup analyses, review authors independently categorised the approach underlying the cognitive intervention as either 'top-down' (interventions that encourage awareness of the disability and potential compensatory strategies) or 'bottom-up' (interventions directed at the impairment but not requiring awareness or behavioural change, e.g. wearing prisms or patches). We included 23 RCTs with 628 participants (adding 11 new RCTs involving 322 new participants for this update). Only 11 studies were assessed to have adequate allocation concealment, and only four studies to have a low risk of bias in all categories assessed. Most studies measured outcomes using standardised neglect assessments: 15 studies measured effect on activities of daily living (ADL) immediately after the end of the intervention period, but only six reported persisting effects on ADL. One study (30 participants) reported discharge destination and one study (eight participants) reported the number of falls.Eighteen of the 23 included RCTs compared cognitive rehabilitation with any control intervention (placebo, attention or no treatment). Meta-analyses demonstrated no statistically significant effect of cognitive rehabilitation, compared with control, for persisting effects on either ADL (five studies, 143 participants) or standardised neglect assessments (eight studies, 172 participants), or for immediate effects on ADL (10 studies, 343 participants). In contrast, we found a statistically significant effect in favour of cognitive rehabilitation compared with control, for immediate effects on standardised neglect assessments (16 studies, 437 participants, standardised mean difference (SMD) 0.35, 95% confidence interval (CI) 0.09 to 0.62). However, sensitivity analyses including only studies of high methodological quality removed evidence of a significant effect of cognitive rehabilitation.Additionally, five of the 23 included RCTs compared one cognitive rehabilitation intervention with another. These included three studies comparing a visual scanning intervention with another cognitive rehabilitation intervention, and two studies (three comparison groups) comparing a visual scanning intervention plus another cognitive rehabilitation intervention with a visual scanning intervention alone. Only two small studies reported a measure of functional disability and there was considerable heterogeneity within these subgroups (I² > 40%) when we pooled standardised neglect assessment data, limiting the ability to draw generalised conclusions.Subgroup analyses exploring the effect of having an attention control demonstrated some evidence of a statistically significant difference between those comparing rehabilitation with attention control and those with another control or no treatment group, for immediate effects on standardised neglect assessments (test for subgroup differences, P = 0.04). The effectiveness of cognitive rehabilitation interventions for reducing the disabling effects of neglect and increasing independence remains unproven. As a consequence, no rehabilitation approach can be supported or refuted based on current evidence from RCTs. However, there is some very limited evidence that cognitive rehabilitation may have an immediate beneficial effect on tests of neglect. This emerging evidence justifies further clinical trials of cognitive rehabilitation for neglect. However, future studies need to have appropriate high quality methodological design and reporting, to examine persisting effects of treatment and to include an attention control comparator.