Waveforms of welding signals such as welding voltage and current include the weld quality. Since uniformity of the waveforms guarantees the stability of the arc and the weld quality, the studies on assessing and controlling the weld quality by monitoring the welding signals have been conducted. In this work, the welding voltage and arc light signal are used to assess the weld quality of the short-circuit gas metal arc welding. The two-dimensional graph of the welding signals and the time derivatives of the signals is divided into subregions by arc characteristics. The weld quality and arc stability are evaluated by the subregion-passing history and the homogeneity of the distribution of points in each subregion. The weld quality index is the function of the weights of the abnormal waveforms and the weight of the short-circuit frequency change. These weights are determined experimentally, and the proposed weld quality assessment method is tested under various welding conditions such as lack of shielding gas, surface contamination and existence of the gap. The results demonstrate some possibility of assessing weld quality in real-time.