OBJECTIVE: The purpose of this systematic review with meta-analysis is to determine the predictive value of the ABCD (2)at 7 and 90 days across three strata of risk. Background. The risk of stroke after transient ischaemic attack (TIA) is significant. The ABCD (2)clinical prediction rule is designed to predict early risk of stroke after TIA. A number of independent validation studies have been conducted since the rule was derived. METHODS: A systematic literature search was conducted to identify studies that validated the ABCD(2). The derived rule was used as a predictive model and applied to subsequent validation studies. Comparisons were made between observed and predicted number of strokes stratified by risk group: low (0-3 points), moderate (4-5 points) and high (6-7 points). Pooled results are presented as risk ratios (RRs) with 95% confidence intervals (CIs), in terms of over-prediction (RR > 1) or under-prediction (RR < 1) of stroke at 7 and 90 days. RESULTS: We include 16 validation studies. Fourteen studies report 7-day stroke risk (n = 6282, 388 strokes). The ABCD(2) rule correctly predicts occurrence of stroke at 7 days across all three risk strata: low [RR 0.86, 95% CI (0.47-1.58), I(2) = 16%], moderate [RR 0.99, 95% CI (0.67-1.47), I(2) = 68%] and high [RR 0.84, 95% CI (0.6-1.19), I(2) = 46%]. Eleven studies report 90-day stroke risk (n = 6304). There is a non-significant trend towards over-prediction of stroke in all risk categories at 90 days. There are 426 strokes observed in contrast to a predicted 626 strokes. As the trichotomized ABCD(2) score increases, the risk of stroke increases (P < 0.01). There is no evidence of publication bias in these studies (P > 0.05). CONCLUSION: The ABCD(2) is a useful CPR, particularly in relation to 7-day risk of stroke.OBJECTIVE: The purpose of this systematic review with meta-analysis is to determine the predictive value of the ABCD (2)at 7 and 90 days across three strata of risk. Background. The risk of stroke after transient ischaemic attack (TIA) is significant. The ABCD (2)clinical prediction rule is designed to predict early risk of stroke after TIA. A number of independent validation studies have been conducted since the rule was derived. METHODS: A systematic literature search was conducted to identify studies that validated the ABCD(2). The derived rule was used as a predictive model and applied to subsequent validation studies. Comparisons were made between observed and predicted number of strokes stratified by risk group: low (0-3 points), moderate (4-5 points) and high (6-7 points). Pooled results are presented as risk ratios (RRs) with 95% confidence intervals (CIs), in terms of over-prediction (RR > 1) or under-prediction (RR < 1) of stroke at 7 and 90 days. RESULTS: We include 16 validation studies. Fourteen studies report 7-day stroke risk (n = 6282, 388 strokes). The ABCD(2) rule correctly predicts occurrence of stroke at 7 days across all three risk strata: low [RR 0.86, 95% CI (0.47-1.58), I(2) = 16%], moderate [RR 0.99, 95% CI (0.67-1.47), I(2) = 68%] and high [RR 0.84, 95% CI (0.6-1.19), I(2) = 46%]. Eleven studies report 90-day stroke risk (n = 6304). There is a non-significant trend towards over-prediction of stroke in all risk categories at 90 days. There are 426 strokes observed in contrast to a predicted 626 strokes. As the trichotomized ABCD(2) score increases, the risk of stroke increases (P < 0.01). There is no evidence of publication bias in these studies (P > 0.05). CONCLUSION: The ABCD(2) is a useful CPR, particularly in relation to 7-day risk of stroke.