To determine how much physical activity, in the form of walking, can be safely and feasibly tolerated for people with severe knee osteoarthritis (OA).Phase I dose response trial with escalating walking doses of 10, 20, 35, 50, 70, and 95 min over 1 week, were prescribed non-randomly to people with severe knee OA. The primary stopping rule was a substantial increase in knee pain. The primary outcomes were an estimation of the maximum tolerated dose of walking; and the proportion of people who did not complete the dose for feasibility reasons. The secondary outcomes were pain, stiffness and activity limitation Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC).Twenty-four participants (13 women) aged 53-83 years, and average body mass index (BMI) of 34 kg/m(2) (SD 9) were recruited. Three participants were assigned to each dose between 10 and 70 min, and nine participants assigned to the 95-min dose. The trial was stopped at 95 min due to the maximum number of adverse events occurring at this dose. Therefore, the maximum tolerated dose was 70 min. No participant stopped due to reasons related to feasibility. There was a moderate association between dose and increased activity (linear R(2) = 0.31, cubic R(2) = 0.69) and reduced stiffness (linear R(2) = 0.20, cubic R(2) = 0.52), with increased benefits at moderate to higher doses.There is preliminary evidence that 70 min per week of moderate intensity supervised walking was safe and feasible for people with severe OA of the knee; for higher doses there was a risk of exacerbating knee pain levels.