We propose that a severe anomaly among the parameters of the gold acceptor state in silicon can be resolved by taking into account the temperature dependence of all the parameters. The temperature variation of most consequence is that of the trap energy level with respect to the band edges. We show that if the trap maintains a constant relative position in the gap, the resulting corrections to the activation energies place the trap nearly 0.1 eV higher in the gap at 300°K than was supposed, and all of the available data are brought into agreement.