Abstract The research purpose of this paper is to estimate the parameter vector (β, σ2,σ2w)of (1/f)-type fractal stochastic processes. Using wavelets, the paper has performed a series of algebraic operation to the variance of the observation wavelet coefficients of process, and presented the elaborate theoretical analysis. As a result, the parameter vector of fractional Brownian motions (fBm) in noise is introduced. The experimental results demonstrate that the new estimator is far simpler and more effective than the traditional ML estimator and the range of estimate parameter is wider. Moreover the distribution of noise is not restricted within Gauss processes.