Different measures of variability of precipitation are discussed and it is concluded that it is preferable to apply therelative intersequential variability in cases when overall trends are feared in the records. 127 stations in Sweden have been studied for the period 1901–1950 as to the relative interannual variability of precipitation. For a selection of these stations other measures have also been calculated and a comparison has been made between the coefficient of variation and the intersequential variability. As the relation between these measures varies with the existence of serial correlations in the records the serial correlation coefficient for a lag of one element has been calculated. The value of the coefficient has indicated the existence of trends in certain areas of Sweden generally in agreement with earlier investigations byAngstrom.