ABSTRACT Aims To obtain preliminary data on changes in gait from the use of a green-lipped mussel (Perna canaliculus) extract product in working farm dogs with musculoskeletal abnormalities using accelerometry. Methods: New Zealand working farm dogs (n = 32) with signs of musculoskeletal abnormalities were enrolled in a double-blinded, placebo-controlled cross-over study. Each dog was allocated to one of six groups to receive three trial substances (180 mg full fat green-lipped mussel extract (GLME180); 220 mg full fat green-lipped mussel extract (GLME220); placebo) in one of the six possible different orders. Each trial substance was administered orally once a day for an 8-week period, with a 4-week washout in between each. Dogs wore a collar-mounted triaxial accelerometer for the study duration. Diet and activity were not controlled. Accelerations were recorded continuously and analysed (n = 27) in 10-second activity epochs partitioned into daytime and night-time periods. Analysis of activity during the daytime period was limited to epochs when dogs were gaiting faster than a walk. The median and IQR of activity were determined for the daytime and night-time. Additionally, the 75th and 90th percentiles of daytime activity for each 24-hour period were determined. Mixed effects linear regression models were constructed to determine if each trial substance altered the response variables. Results During the daytime, the 90th percentile was higher when dogs were given GLME220 compared with the placebo (β coefficient 2.6; 95% CI = 0.25–4.94; p = 0.03). Dogs that started the trial with the GLME products had a higher 90th percentile activity compared with dogs that began with the placebo (β coefficient 26.26; 95% CI = 0.45–52.06; p = 0.046). The 75th percentile for activity was not affected by the GLME product. The daytime IQR was larger when dogs were given the GLME180 product compared with the placebo (β coefficient 1.25; 95% CI = 0.12–2.37; p = 0.03). Night-time median activity and the IQR was greater in dogs that started the trial with the GLME products than in dogs that began with the placebo. The night-time IQR for activity was greater for GLME180 than for the placebo. Conclusions Administration of a low dose of the GLME-containing product increased peak activity in working farm dogs with signs of musculoskeletal abnormalities and may improve their performance. Clinical relevance Even mildly affected working farm dogs might benefit from support of their musculoskeletal abnormalities, and this particular GLME-based product shows promise as an adjunct to other management strategies.