Nondestructive techniques for appraising gas metal arc welding(GMAW) faults plays a very important role in on-line quality controllability and prediction of the GMAW process. On-line welding quality controllability and prediction have several disadvantages such as high cost, low efficiency, complication and greatly being affected by the environment. An enhanced, efficient evaluation technique for evaluating welding faults based on Mahalanobis distance(MD) and normal distribution is presented. In addition, a new piece of equipment, designated the weld quality tester(WQT), is developed based on the proposed evaluation technique. MD is superior to other multidimensional distances such as Euclidean distance because the covariance matrix used for calculating MD takes into account correlations in the data and scaling. The values of MD obtained from welding current and arc voltage are assumed to follow a normal distribution. The normal distribution has two parameters: the mean ? and standard deviation of the data. In the proposed evaluation technique used by the WQT, values of MD located in the range from zero to ?+3? are regarded as “good”. Two experiments which involve changing the flow of shielding gas and smearing paint on the surface of the substrate are conducted in order to verify the sensitivity of the proposed evaluation technique and the feasibility of using WQT. The experimental results demonstrate the usefulness of the WQT for evaluating welding quality. The proposed technique can be applied to implement the on-line welding quality controllability and prediction, which is of great importance to design some novel equipment for weld quality detection. Nondestructive techniques for appraising gas metal arc welding (GMAW) faults plays a very important role in on-line quality controllability and prediction of the GMAW process. On-line welding quality controllability and prediction have several advantages such as high cost, low efficiency, complication An enhanced, efficient evaluation technique for evaluating welding faults based on Mahalanobis distance (MD) and normal distribution is presented. In addition, a new piece of equipment, designated the weld quality tester (WQT), is developed based on the proposed evaluation technique. MD is superior to other multidimensional distances such as Euclidean distance because the covariance matrix used for computing MD takes into account correlations in the data and scaling. The values ​​of MD obtained from welding current and arc voltage are known to follow a normal distribution. The normal distribution has two parameters: the mean? and standard deviation of the data. In the proposed evaluation technique used by the WQT, values ​​of MD located in the range from zero to? +3? are as as “good”. Two experiments which involve changing the flow of shielding gas and smearing paint on the surface of the substrate are conducted in order to verify the sensitivity of the proposed evaluation technique and the feasibility of using WQT. The experimental results demonstrate the usefulness of the WQT for evaluating welding quality. The proposed technique can be applied to implement the on-line welding quality controllability and prediction, which is of great importance to design some novel equipment for weld quality detection.