The sputtering yield of copper has been measured with 21 different 45-keV ions as projectiles. The measurements were performed by means of an oscillating quartz micro-balance. This allows the sputtering yield to be accurately measured even if only 40–50 Å of the target material was sputtered away. It was found that the measured sputtering yield depended sensitively on the previous irradiation of the target. We therefore preferred to measure all sputtering yields relative to copper self-sputtering. This ratio of sputtering yields was found to be insensitive to previous bombardments and to agree well with Sigmund's theory, although small systematic disagreements are found. The overall agreement implies firstly that the modifications of the target during bombardment only affect the ejection conditions at the surface, and secondly that any Z1oscillations of the nuclear stopping power are less than 10 per cent.