This correspondence presents a general, computationally efficient method fu
r chaotic signal estimation based on the connection between the symbolic se
quence and the initial rendition of a chaotic system. The performance of th
e method in white Gaussian noise is evaluated, The new method is asymptotic
ally unbiased and attains the Cramer-Rao lower bound at high signal-to-nois
e ratios.