Abstract. The LARES satellite, a laser-ranged space experiment to contribute to geophysics observation, and to measure the general relativistic Lense-Thirring effect, has been observed to undergo an anomalous along-track orbital acceleration of -0.4 pm/s 2 (pm := picometer). This thermal "drag" is not surprising; along track thermal drag has previously been observed with the related LAGEOS satellites (-3.4 pm/s 2 ). It is hypothesized that the thermal drag is principally due to anisotropic thermal radiation from the satellite's exterior. We report the results of numerical computations of the along-track orbital decay of the LARES satellite during the first 126 days after launch. The results depend to a significant degree on the visual and IR absorbance α and emissivity of the fused silica Cube Corner Reflectors. We present results for two values of αIR = IR: 0.82, a standard number for "clean" fused silica; and 0.60, a possible value for silica with slight surface contamination subjected to the space environment. The heating and the resultant along-track acceleration depend on the plane of the orbit, the sun position, and in particular on the occurrence of eclipses, all of which are functions of time. Thus we compute the thermal drag for specific days. We compare our model to observational data, available for a 120-day period starting with the 7th day after launch, which shows the average acceleration of -0.4 pm/s 2 . With our model the average along-track thermal drag over this 120-day period for CCR αIR = IR = 0.82 was computed to be -0.59 pm/s 2 . For CCR αIR = IR = 0.60 we compute -0.36 pm/s 2 . LARES consists of a solid spherical tungsten sphere, into which the CCRs are set in colatitude circles. Our calculation models the satellite as 93 isothermal elements: the tungsten part, and each of the 92 Cube Corner Reflectors. The satellite is heated from two sources: sunlight and Earth's infrared (IR) radiation. We work in the fast-spin regime, where CCRs with the same colatitude can be taken to have the same temperature. Since all temperature variations (temporal or spatial) are expected to be small, we linearize the Stefan-Boltzmann law and, taking advantage of the linearity, we perform a Fourier series analysis. The variations are indeed small, validating our Fourier analysis.