A simple model of the crash phase of sawteeth in tokamaks is presented, which explains many of the significant experimental observations: small oscillations in the electron temperature prior to the crash (precursor), a short crash time, and large long‐lived oscillations in the electron temperature following the crash (successor). In tokamak discharges with good central confinement the temperature profile across the center of the discharge remains rather flat. In such profiles skin currents form that cause the safety factorqto fall below one at a finite radiusrswhileq(r=0)&bartil;1. The sawtooth crash is caused by them=n=1 tearing mode growing on such a profile. For a profile in whichqfalls below one by an amount &Dgr;q, the crash time is given by the helical Alfve´n time &tgr;A/&Dgr;q, independent of the resisitivity. The size of the magnetic island of the precursor can be a small fraction ofrs. In contrast, a remnant magnetic island of the order ofrssurvives the crash and causes the successor oscillation. Estimates of the decay time of the successor are given.