Sedentary behaviour is a major risk factor for chronic disease morbidity and mortality in aging. Measuring people activity through devices such as pedometers is a recognized intervention to motivate them for more physical activity. However, the feedback provided by these devices must be accurate in order to avoid overtraining and keep users' motivation alive. If the accuracy of pedometers has been validated for healthy people, their lack of accuracy for elderly people walking at slower pace has been reported in several studies. The emergence on the consumer's market of new devices that can be worn indifferently at the wrist or at the waist raises once more this concern. In order to evaluate whether pedometers' location influences their accuracy, we have tested three pedometers at different locations, and for several paces in a comparative study. Beyond confirming the decrease of pedometers' accuracy with speed reduction, our study reveals that pedometers should be worn at the waist rather than at the wrist. This leads us to recommend wearing pedometers at the waist when monitoring population with reduced mobility.