Biomarkers are required to track disease progression and measure the effectiveness of interventions for people with spinocerebellar ataxia type-6 (SCA6). Gait is a potential biomarker that is sensitive to SCA6 which can be measured using wearable technology, reducing the need for expensive specialist facilities. However, algorithms used to calculate gait using data from wearables have not been validated in SCA6. This study sought to examine the validity of a single wearable for deriving 14 spatio-temporal gait characteristics in SCA6 and control cohorts. Participants performed eight intermittent walks along a 7 m instrumented walkway at their preferred walking pace while also wearing a single accelerometer-based wearable on L5. Gait algorithms previously validated in neurological populations and controls were used to derive gait characteristics. We assessed the bias, agreement and sensitivity of gait characteristics derived using the instrumented walkway and the wearable. Mean gait characteristics showed good to excellent agreement for both groups, although gait variability and asymmetry showed poor agreement between the two systems. Agreement improved considerably in the SCA6 group when people who used walking sticks were excluded from the analysis, suggesting poorer agreement in people with more severe gait impairment. Despite poor agreement for some characteristics, gait measured using the wearable was generally more sensitive to group differences than the instrumented walkway. Our findings indicate mean gait characteristics can be accurately measured using an accelerometer-based wearable in people SCA6 with mild-to-moderately severe gait impairment yet further development of algorithms are required for people with more severe symptoms.