The principal concern in the calibration of leaks is the calibration of leaks to be used under vacuum. With a growing number of sniffing applications and the change from permeation leak to physical leak elements due to reasons of stability and reproducibility, the study focuses on the calibration of physical leaks under application conditions. The calibration of these test leaks must be carried out at atmospheric pressure. To get an accurate calibration independent of the tilling gas for these sniffing test leaks a P Delta V/Delta t calibration system was built according to DIN28417. The linear displacement of a piston gives the change in volume, Delta V. The smallest Delta V that can be set is 3 nl, which corresponds, at a pressure of 1000 mbar, to 3x10(-6) mbar 1. The systematic errors in Delta V are determined by the diameter and tolerances of the piston, 0.6+/-0.0015 mm or 0.25%, and the measurement of the displacement of the piston h and its error Delta h, which could be maintained at less than 0.01 mm or 2% for h=0.5 mm. The overall systematic errors are below 2.5%. With this leak calibration apparatus, sniffing test leaks with leak rates of 2x10(-5) mbar 1/s at room temperature can be calibrated with an accuracy of better than +/-10% within several minutes. In such a calibration the Delta V is performed 20 times to get 20 individual determinations of the leak rate. The comparison of this method with a capillary measurement tube method shows very good agreement.