Objective:To study extracellular multi-neuron activity in the nervous system based on wavelet-fractal technique. Methods: The wavelet transform was employed to decompose the original signal and obtain 4 sub-patts. The dividing fractal dimensions of these sub-patts were computed. A knnclassier was used to classify feature vectors. Results:Not all the elements in feature vector DimDC were very powerful for this patt recognition problem through the empirical study of noise signals. The most effective feature vector was defined as DimDC = (d3; d4) above. Conclusion: Wavelet-fractal algorithm has high accuracy and provides a powerful tool for clinical application.