Detailed examination of a large number of copper water tanks and piper showing pitting corrosion (taken from certain districts to which pitting is confined) revealed features that could not be reconciled with the conventional explanation based on the presence of large cathodic areas surrounding small anodic corrosion sites. A new theory of pitting corrosion of copper is proposed which explains the features observed. Pitting arises when a pocket of cuprous chloride forms beneath a porous electrically-conducting membrane (usually cuprous oxide). The anodic and cathodic processes occur on the inner and outer surfaces of the membrane. Dissolution of copper occurs by reaction with the anodic product beneath the membrane and deposition of calcium carbonate occurs by reaction of hardness salts in the water with the cathodic product above the membrane. Pockets of cuprous chloride capable of developing into corrosion pits can only form if the potential of the copper rises above 90 mV E_H during the period of film formation. Control of the potential below this value by coupling to a small sacrificial aluminium anode has been shown to prevent formation of pits in copper tanks and cylinders in service.