We present a detailed numerical analysis to quantify the power generation performance of a thermoelectric module in radiant heat recovery application. Due to the large temperature difference typically involved in such a system, temperature-dependent material properties of thermoelectric elements are taken into account for accurate performance prediction by employing an iterative algorithm based on the one-dimensional finite element method. Careful analysis on the radiation heat transfer with optical parameters such as surface emissivity and view factor is performed to precisely quantify the heat input to the thermoelectric system. Parasitic heat losses such as air convection loss at the hot surface and conduction through the substrates and gap fillers are also taken into account to analyze their impacts on the power output. A case study on the radiant waste heat recovery from hot steel casting slabs in steel industry is discussed in detail to theoretically estimate the power output performances and optimize the module design. We find that a power density as high as similar to 1.5 kW/m(2) and a system efficiency as high as similar to 4.6% can be achieved at a 2 m distance from the 1200 K hot steel slab using the state-of-the-art Bi2Te3 alloys with a relatively small leg thickness of 3 mm and a 20% fill factor. This optimal design with small form factors ensures a reduced material cost while keeping the power output near the maximum, so that an estimated power cost remains as low as similar to 0.2$/Watt.