We investigate various possible definitions of an effective temperature for a particularly simple nonequilibrium stationary system, namely a heated Brownian particle suspended in a fluid. The effective temperature based on the fluctuation dissipation
ratio depends on the time scale under consideration, so that a simple Langevin description of the heated particle is impossible. The short and long time limits of this effective temperature are shown to be consistent with the temperatures estimated from the kinetic energy and Einstein relation, respectively. The fluctuation theorem provides still another definition of the temperature, which is shown to coincide with the short time value of the fluctuation dissipation ratio.
