This paper introduced an enhancement version of Quality Threshold ARTMAP using the Mahalanobis function. It’s known as Mahalanobis Quality Threshold ARTMAP (QTAM-m) that increase its capability for pattern classification and prediction purposes. In addition this enhancement also does not consist any initial parameters setting that will affect the final classification outcomes. Thus it’s fulfilling the requirement of on-line learning scheme. All parameters involved are also updated using an equation that produced the exact value and not just based on estimation. In general the results had indicates that this improvement increase the classification result based on several testing of benchmark dataset. The proposed technique is also compared with several other techniques within it class.