This paper presents a novel frequency distribution scheme intended to provide more accurate estimations of Cole parameters. Nowadays a logarithmic frequency distribution is mostly used in Electrical Bioimpedance Spectroscopy (EBIS) applications. However it is not optimized following any criterion. Our hypothesis is that an EBIS signal contains more information where the variation of the measurement regarding the frequency is larger; and that there ought to be more measuring frequencies where there is more information. Results show that for EBIS data with characteristic frequencies up to 200 kHz the error obtained with both frequency distribution schemes is similar. However, for EBIS data with higher values of characteristic frequency the error produced when estimating the values from EBIS measurements using an adaptive frequency distribution is smaller. Thus it may usefull for EBIS applications with high values of characteristic frequency, e.g. cerebral bioimpedance.