Remote ischaemic conditioning (RIC) is achieved by repeated transient ischaemia of a distant organ/limb and is neuroprotective in experimental ischaemic stroke. However, the optimal time and methods of administration are unclear. Systematic review identified relevant preclinical studies; two authors independently extracted data on infarct volume, neurological deficit, RIC method (administration time, site, cycle number, length of limb occlusion (dose)), species and quality. Data were analysed using random effects models; results expressed as standardised mean difference (SMD). In 57 publications incorporating 99 experiments (1406 rats, 101 mice, 14 monkeys), RIC reduced lesion volume in transient (SMD −2.0; 95% CI −2.38, −1.61; p < 0.00001) and permanent (SMD −1.54; 95% CI −2.38, −1.61; p < 0.00001) focal models of ischaemia and improved neurological deficit (SMD −1.63; 95% CI −1.97, −1.29, p < 0.00001). In meta-regression, cycle length and number, dose and limb number did not interact with infarct volume, although country and physiological monitoring during anaesthesia did. In all studies, RIC was ineffective if the dose was <10 or ≥50 min. Median study quality was 7 (range 4–9/10); Egger’s test suggested publication bias ( p < 0.001). RIC is most effective in experimental stroke using a dose between 10 and 45 min. Further studies using repeated dosing in animals with co-morbidities are warranted.