Objective To analyze human gait stability by acceleration signal at the head and lumbar under different walking conditions, and make comparison with parameters by the traditional COM (center of mass)-COP(center of pressure) method, so as to discuss the reliability of applying wearable sensors to analyze human gait stability. Methods The harmonic ratio (HR) parameter at the head and lumbar based on acceleration signal was applied to analyze gait stability of 18 healthy young adults under 3 walking conditions (footwear normal walking, barefoot normal walking and barefoot walking at different velocities), and the results were compared with the assessment results from the COM-COP method. Results Walking at normal velocity was most stable, with the maximum HR parameter. Compared with footwear walking, HR parameters were significantly decreased (P＜0.05) during barefoot walking, indicating that gait stability was reduced. The results were consistent with the assessment results from the COM-COP method. Considering the factors of walking velocity and footwear, the gait stability parameters obtained by the two methods showed a significant negative linear correlation (R2＞0.50). Lumbar HR parameter and COM-COP parameters showed a stronger linear correlation (R2＞0.65). Conclusions The application of acceleration signal-based analysis algorithm could effectively and reliably evaluate the stability of human gait, and acceleration at the lumbar was more sensitive than the head signal for analyzing gait stability.