Applications in geodesy and engineering surveying require the determination of the heights of the vertical control points in the national and local networks using different techniques. These techniques can be classified as geometric, trigonometric, barometric and Global Positioning System (GPS) levelling. The aim of this study is to analyse height differences obtained from these three techniques using precise digital level and digital level, total station (trigonometric levelling) and GPS which collects phase and code observations (GPS levelling). The accuracies of these methods are analysed. The results obtained show that the precise digital levelling is more stable and reliable than the other two methods. The results of the three levelling methods agree with each other within a few millimetres. The different levelling methods are compared. Geometric levelling is usually accepted as being more accurate than the other methods. The discrepancy between geometric levelling and short range trigonometric levelling is at the level of 8 millimetres. The accuracy of the short range trigonometric levelling is due the reciprocal and simultaneous observations of the zenith angles and slope distances over relative short distances of 250 m. The difference between the ellipsoidal height differences obtained from the GPS levelling used without geoid and the orthometric height differences obtained from precise geometric levelling is 4 millimetres. The geoid model which is obtained from a fifth order polynomial fit of the project area is good enough in this study. The discrepancy between the precise geometric and GPS levelling (with geoid corrections) is 4 millimetres over 5 km.