Autocorrelation and spectrum analyses of amino acid residues along protein chains in a large data base has been performed. Results reveal the presence of general long range correlations. Similar analyses of simulated (random) peptides do not exhibit any such long range correlations. Based on the results of nur analysis, an attempt has been made to model the distribution of residues in protein sequences on a fractional Brownian motion and individual sequences as multi-fractals. For this purpose, the characteristics of an fractional Brownian motion namely, the scaling parameterH. the spectral exponent β and the fractal dimensionD, have been described