For a study, researchers sought to develop and test a method for step detection using accelerometer data in individuals with lumbar spinal stenosis (LSS). There were 2 objectives:  to describe a method for identifying steps using accelerations felt at the wrist, hip, lower back, thigh, and ankle; and to assess the method’s precision when using walking aids, when using none at all, and when performing non-walking tasks. About 30 people with LSS participated in a standardized mobility protocol that involved walking with and without canes and wearing accelerometers at 5 different wear sites while doing non-walking activities. Following the walking tests, the best step detection method was developed, and it was then evaluated against the gold standard of real step counts. The step identification technique applied to accelerations from the lower back, hip, thigh, and ankle during continuous walking without using walking devices produced accurate step counts. With the exception of the ankle, accuracy suffered when using a walking aid at all wear-sites. The wrist produced the most erroneous step count, while the accelerometers on the thigh and ankle were prone to making mistakes while riding. The ankle-worn accelerometer gave the most accurate step count. However, non-walking activities were miscounted as steps. The developed step detection method shows potential as a gauge of walking activity, highlighting the need for additional study and testing in real-world settings.