To describe how physical activity is monitored following stroke; to summarise methods and devices used across the stroke pathway and document their psychometric properties.Searches of five databases identified studies that included stroke survivors whose physical activity was quantitatively measured. Two reviewers independently determined inclusion. A descriptive synthesis was undertaken and reliability data for specific methods of monitoring physical activity were pooled where possible.Ninety-one papers (60 using devices and 31 using observational methods) met inclusion criteria, with 3479 participants aged 21-96 years. Twenty-nine devices (72% accelerometers) were identified. Devices were typically used to measure ambulant participants more than 6 months following stroke. Direct observation of physical activity was commonly used for inpatients. No outcome measurements were common to all methods/devices. Test-retest reliability was not reported for 23 devices; for the remaining six it ranged from r = 0.44 to r = 0.99. Inter-rater reliability of observational methods ranged from 0.51 to 1.0. Validity was infrequently reported.Physical activity outcomes were variable. Devices allow for unobtrusive, sustained monitoring in free-living environments. Observational methods suit inpatient settings but are time and labour intensive. No single approach appears superior but standardisation of outcomes would improve the field.Physical activity measurement is highly variable following stroke and better definition of physical activity outcomes would enhance the field. Accelerometry and behavioural mapping are most commonly used to measure physical activity following stroke, each have advantages and disadvantages depending on the setting and the outcome of interest. There is no single device ideal for clinical application for people following stroke.