K e y w o r d s: levelling networks, random errors, systematic errors. A b s t r a c t Paper presents statistical evaluation of accuracy of levelling network measured in Poland in years 1999–2001. The analysis was done using 16 150 misclosures from the double levelling of the sections, 382 misclosures from the double levelling of the lines and 133 loops misclosures. The statistical analysis was conducted by the regression method, correlation method and the analysis of variance. It results that the measured height differences have various accuracy (analysis of variance), and that systematic errors are changing according to the value and sign. The existence of systematic errors causes that the successive neighboring sections of some levelling...