Detecting the presence of liquid slag in the pouring or tapping stream of a basic oxygen furnace after steel-making offers a chance to aid the process by maximizing the amount of steel removed from the vessel while minimizing the amount of slag deposited into the holding vessel or ladle. Thermal imaging has been shown capable of real-time images that easily discriminate between the two materials by exploiting their differences in apparent temperature. However, the underlying radiation thermometry theory has yet to be published. Combine this lack of theory with the need for very robust equipment to provide 24/7 uptime and serious obstacles to equipment investment appear. This presentation reviews some basic radiation thermometry theory, along with material properties to show that significant discrimination differences between slag and steel exist in different wavelength passbands. Combining these differences with the imaging, lifetime and cost properties of commercially available devices enables one to develop the necessary tradeoffs between discrimination ability, image quality, equipment reliability and system costs.