The calculation of the statistical probability that a flood of a certain magnitude for a given river will occur in a certain period of time. Each flood of the river is recorded and ranked in order of magnitude with the highest rank being assigned to the largest flood. The return period here is the likely time interval between floods of a given magnitude and can be calculated as:

All methods of FFA are methods of extrapolation, which requires the fitting of a model, and this, in turn, needs an assumption about the underlying distribution generating flood events: ‘not only is this not known for extreme hydrological events beyond the observed record, but it is untestable within human timescales’ (Kidson and Richards (2005) PPG 29, 3).