This paper presents the surface energy balance at five locations on the Pasterze, a glacier in Austria, during a period of 46 days in the summer of 1994. The computations are based on local measurements of radiative fluxes, wind speed, atmospheric temperature and humidity, and precipitation. These measurements served as input for a numerical mass‐balance model that computes the surface temperature by simulation of subsurface processes. The surface temperature was then used for the computation of the outgoing long‐wave radiative flux and for the computation of the turbulent fluxes by means of the “bulk method.” Values of the roughness length for momentum were computed by profile analysis. They ranged from 1.2 to 5.8 mm. At all locations, net short‐wave radiation constituted the dominant energy flux, the latent heat flux was small, and the net long‐wave radiative flux and the sensible heat flux were of intermediate magnitude. The energy balances at the three stations on the glacier tongue were similar to each other but differed from those at the two stations in the accumulation basin, which were also similar to each other. These similarities within the two areas are due mainly to the fact that during the experiment there was hardly any variation in temperature and albedo within the two areas. The ablation rate at U3 (2420 m above sea level (asl)) was even slightly higher than the ablation rate at Al (2205 m asl). At all stations the calculated mass balance agrees within 5% with the measured mass balance, which is within the margins determined by the uncertainties in the input variables and the model parameters and equations. This gives some confidence in the method used to compute the turbulent fluxes, but on the other hand, the agreement may also be caused by cancellation of errors. Because of the high temperatures during PASTEX the calculated mass balance would have been almost identical if the “zero‐degree assumption” had been used instead of the subsurface module.