Background: therapists and nurses often use verbal instruction in the rehabilitation of mobility following stroke. This study aimed to determine whether performing a verbal cognitive task while walking adversely affected patients' balance and velocity.Methods: there were two counterbalanced conditions: walking only and walking and concurrent cognitive activity. The cognitive activity used was to give one of two verbal responses to two verbal stimuli. An electronic GaitMat measured gait velocity and balance (double support time as a percentage of stride time).Results: 11 people with stroke participated in the study (five women and six men, mean age 72 years, SD 9). They were on average 120 (SD 48) days post-stroke. Velocity decreased (P=0.017) and double-support time as a percentage of stride time increased (P=0.010) when the cognitive activity was added to the test.Conclusions: performing a verbal cognitive task while walking adversely affected stroke patients' balance and gait velocity. Susceptibility to disruption within the patient group, suggesting clinical heterogeneity. Further research is required before changes to clinical practice are justified.