Possibility of increasing the precision of quantitative determination of memantine by capillary electrophoresis (CE) was studied using artificial neural networks (ANNs). ANNs show their unique merits in the calibration of non-linear models. In the analysis of memantine by CE, there is no obvious linear relationships between the concentrations of the memantine and its peak areas or heights, or even their ratios to those of the corresponding internal standards, respectively. The greatest advantage of ANNs is that no prior knowledge of migration behavior and separation system is needed. The inputs of the ANNs were the peak areas and heights of memantine, and the outputs were the concentrations of the memantine. The 2:2:1 ANNs gave excellent results. The ANNs approach can also be used to improve the precision of determination of other kinds of analytes for its general validity.

Key words: CAPILLARY ZONE ELECTROPHORESIS, DRUGS ANALYSIS