As instrumental records often cover only short time periods, historical information is the main source of data in order to extend natural disaster catalogues. This study aims to show the feasibility of quantifying extreme sea levels and skew surges during storm events by using historic documentary data. The documentary data related to the extreme events is often not freestanding enough to directly extract water levels and subsequently skew surge levels, hence auxiliary information such as dike sketches was used to interpret the collected data. First a strategy for the reconstruction of historic levels is presented which implies an analysis on three different scales: (i) the data scale, (ii) the quantification scale and (iii) the event scale. Strong hypotheses were made during the quantification processes of the seven historic skew surges. Therefore the methodology also aims to trace the hypotheses taken for the surges reconstruction, to inform potential users about the different degrees of reliability of estimated values. Secondly this strategy is applied on an existing database of storms and storm surges that occurred in Dunkirk in the north of France, which contains 75 events. Within this database the focus was set on seven surge events that hit Dunkirk between 1778 and 1846 and seven new skew surges were estimated for that period.