This paper present a method for predicting voltage dips as experienced by customers due to faults in the power grid. Where earlier methods considered only the magnitude (residual voltage) of the dips, this paper considers both magnitude and duration. A triangular distribution is assumed for the dip duration for each of three zones, together with probabilities for each zone. The dips are generated randomly by means of a Monte-Carlo simulation, with fault location along the line and dip duration as random variables. The results can next be analyzed by any of the methods developed for analyzing voltage-dip measurements. The method is illustrated by means of the dips experienced by an industrial installation connected to the 130-kV subtransmission network.