Experiments on surface breakdown at reverse‐biased germaniumn+palloyed junctions are described, and the results related to a new model for surface breakdown that takes explicit account of semiconductor surface charge and fringing field. The experiments show (i) that surface breakdown, like body breakdown, is an avalanche process; (ii) that multiplication sets in first at a particular spot. The experiments serve also to confirm conclusions already reached from device experience: (i) that high breakdown voltage is promoted by that sign of surface charge which tends towards the formation of a ``channel'' over the material of the higher resistivity side, and low breakdown voltage by the opposite sign of surface charge; (ii) that the breakdown voltage is increased by surrounding the material by a medium of high dielectric constant. The theory accounts for all these observations and leads, when the calculations are carried out for an equivalent one‐dimensional structure, to a quantitative prediction of the relation between breakdown voltage, surface charge, resistivity, and the dielectric constant of the surrounding medium. The evidence for the validity of this relation is discussed briefly.