"Electrochemical noise" precedes the onset of stable pitting corrosion of stainless steel in chloride solutions. In this work, micro-electrodes have been used to capture single current noise events, which result from metastable pit growth on Type 304 stainless steel in chloride solutions. The form of the current transients has been studied in some detail, and the rate of occurrence of metastable pits has also been characterised as a function of potential, solution composition, and surface fmish. Stable pit growth requires the presence inside the pit of an acidic solution with a high chloride concentration; it is this aggressive solution which sustains rapid dissolution. This aggressive solution is formed by the hydrolysis of the metal cations which are produced by dissolution of the metal, and migration of chloride ions into the pit. Diffusion tends to dilute the pit anolyte, and a balance between the rate of dissolution (current density), and the rate of diffusion, is required to maintain the anolyte at an aggressive concentration. The rate of diffusion from concave hemispherical pits has been quantified. This analysis indicates that open hemispherical pits can only grow stably if the product of current density and pit depth is greater than a minimum value. This minimum product is not achieved in metastable pits - this implies that a (perforated) cover must be present over the pit mouth, to act as a barrier to diffusion. It is thought that the repassivation of metastably growing pits is caused by the rupture - due to the effect of osmotic pressure - of this cover. The experimentally-observed transition from metastable growth to stable propagation agrees with the predictions based on diffusion calculations. Experimental evidence indicates that metastable pit growth is under diffusion control, i.e independent of the electrode potential. The potential dependent nature of pitting is thought to originate with the pit nucleation process, which is considered to be fundamentally different from the pit growth process.