Temperature–time histories of burning single coal particles can be obtained with multi-color (multi-wavelength) optical pyrometry. With this method, a number of different temperatures can be deduced from the resulting number of two-color ratios. However, these two-color temperatures do not always agree, causing considerable uncertainty in the temperature measurement. This work used a three-color pyrometer and focused on identifying and minimizing the causes of disparity among the three deduced temperatures. Components of the pyrometer (such as dichroic filters, interference filters and photo-detectors) were modeled mathematically, taking into account their wavelength-dependent properties. The pyrometer was calibrated with both a high-temperature pre-calibrated tungsten lamp, and a moderately-high temperature blackbody cavity, to span the temperature range of interest in pulverized coal combustion. Temperatures were deduced based not only on a suitably-modified pyrometric signal ratio method but also, on a similarly modified pyrometric signal non-linear least-square method, to provide comparison. Results are exemplified by presenting radiation-signal-time and temperature–time profiles of single particles burning in air. The variation of the projected luminous area of burning particles was also computed using both methods, and area–time profiles are presented herein. The char particle emissivity was either treated as a quantity independent of the wavelength (i.e., assuming gray-body behavior), or as a quantity assumed to depend linearly on the wavelength and using pertinent published emissivity data. Finally, a sensitivity analysis was performed to investigate individual effects of parameters, such as the calibration method, the wavelength dependencies of filter transmissivities, and the photo-detector responsivities on the pyrometric signal ratio method temperature consistency.