An investigation was made to evaluate the capacity of a colorimetric artificial nose to detect toxic gas at low concentration. A low-cost and simple colorimetric sensor array for identification and quantification of NH3 with different concentrations (30, 90, 150, and 210 ppb) were reported. Using porphyrin, porphyrin derivatives (mainly metalloporphyrins), and chemically responsive dyes as the sensing elements, the developed sensor array of artificial nose showed a unique pattern of colorific change upon its exposure to NH3 with different concentrations. The dynamic responses of colorimetric sensor array to NH3 and colorimetric sensor array to various NH3 concentrations at the same time point showed that there was a positive relationship between the color change values of spots and contractions of NH3. NH3 with four concentrations were measured, and the response values at six different collection times were conducted by linear discrimination analysis (LDA) and artificial neural network (ANN). The four concentrations were discriminated completely by LDA. The response value of the colorimetric artificial nose at 0.4 min was optimum for discrimination. The method of ANN was performed and less than 5% of error by using T-S fuzzy neural network.