The objective of this study is to compare the gait variability of patients with lumbar spinal stenosis (experimental group) with healthy individuals (control group). The hypothesis is that the preoperative gait variability of the experimental group is higher than the control group. The experimental group consisted of 35 adults (18 males, 17 females). The subjects of the experimental group suffered exclusively from spinal stenosis. The patients were determined by MRI scans. A tri-axial accelerometer sensor was used for the gait measurement, and differential entropy algorithm was used to quantify the gait acceleration signal. The Oswestry Low Back Pain Questionnaire was used to determine the condition on the day of the measurement. Receiver operating characteristic (ROC) was utilized to assess the diagnostic value of the method and determine a cut-off value. There is a statistically significant difference between gait variability in the control group and the experimental group. ROC analysis determines a cut-off differential entropy value. The cut-off value has a 97.6% probability of separating patients with spinal stenosis from healthy subjects. The Oswestry Low Back Questionnaire is well correlated with the spectral differential entropy values.