OBJECTIVE To determine the test-retest reliability of the GAITRite system in measuring spatio-temporal gait parameters in people with stroke undergoing inpatient rehabilitation and to investigate whether walking ability influences test-retest reliability. METHODS A within-subject repeated measures study was conducted. Twenty-one individuals with first stroke undergoing inpatient rehabilitation participated. The Motor Assessment Scale Item 5 score was used to categorise participants into poorer (score 3 or 4) or better walkers (score 5 or 6). Participants made two passes at a self-selected gait speed over the GAITRite mat on two separate occasions. Intra-class correlation coefficients (ICC(2,1)) and 95% limits of agreement for spatio-temporal gait parameters between the two occasions and groups were determined. RESULTS No differences in gait parameters between the two sessions (p > 0.15) were found. Test-retest reliability for the whole group was good to excellent, ICC(2,1) ranging from 0.72 to 0.94. Walking ability did influence these findings with fair to good reliability demonstrated for both poorer (ICC(2,1)  >  0.57) and better walkers (ICC(2,1)  >  0.41). CONCLUSIONS It appears that the GAITRite has good test-retest reliability in this group of people with first stroke undergoing rehabilitation. Test reliability seems to be slightly higher for poorer walkers than better walkers.