The slip-dissolution model of stress corrosion cracking states that crack tip advance is based on active dissolution of metal atoms after film rupture and until fill repassivation has occurred. The repassivation phenomenon can be investigated using a scratch test. After a stainless steel sample is polarized in the passive part of the anodic polarization curve, a sharp pin makes a scratch on the surface, hereby removing the passive film. Repassivation of the bared metal in the scratch will occur due to the applied potential. The current supplied to the metal sample is measured before, during and after scratching. A device has been designed that can perform such a test in an autoclave. A ceramic pin produces a scratch on a stainless steel sample when a metal bellow expands under a burst internal pressure. Typical problems encountered include the well controlled movement of the ceramic pins, scratch reproducibility and proper current measurement. Furthermore, a finite scratching speed results in mixed activation/repassivation, i.e. repassivation starts before the scratch is finished. A convolution calculation has been made to tackle this problem. The paper discusses the design and experimental procedure of the scratch test and some experimental results obtained in high temperature water (300degreesC). Mathematical modeling of the repassivation current shows good correlation with experimental data. The mathematical model allows for a slow scratching speed and circumvents some of the disadvantages of conventionally fast scratch tests. (C) 2004 Elsevier Ltd. All rights reserved.