A significant number of research groups is actively investigating the development of so-called "Brain-Computer Interface (BCI)" systems. There are various methods described for good determination and classification of the EEG signals which offer many exciting possibilities for the control of peripheral devices via computer analysis. Most effort has also been concentrated in the analysis of changes in the frequency content and has been carried out using the complexity measures of EEG signals. This work presents a low cost Brain computer Interface using Short Time Fractal Dimension and Wavelets Transforms for signals recorded from single channel EEG using only three electrodes, a single isolation amplifier and a 166 MHz PC. Wavelet transforms were used to decompose the EEG signal into six scaling components (D1, D2, ..., D6).For each decomposition the fractal dimension parameter was computed. The fractal dimension parameter can be computer by means of several methods. The most popular are the correlation dimensions and the box counting methods. This approach is based on the curve covering at different scales. A basic element of size e is selected. By increasing e and computing the corresponding area of the cover, area(e), a number of couples (e, area(e)) is obtained. A straight line is then fitted, using the least square method, obtaining the graph of log[area(e)/e2] versus log(1/e). The approximate estimate of the fractal dimension is the slope of this line. The Fractal dimension of each decomposition constituted the selected features of the EEG signals. The classification is then made by a three layer artificial neural network (ANN) using the fractal dimensions as input parameters. Innovative Software Algorithms