In 2003, Canadian Surface Ocean LowerAtmosphere Study conducted 3 research expeditions (April-May, July and October) to the Northwest Atlantic to study gas exchange at the air-sea interface. Interpreting the relationships between gas exchange and biological activity measured in these expeditions is facilitated if the water masses sampled can be identified in a large-scale context. Longhurst et al. (1995) tackled the issue of water mass characterisation by defining biogeochemical provinces for the world ocean and in particular for the NW Atlantic. To avoid the limitations that arise from the static arrangement of provinces with rectilinear boundaries as used in Longhurst et al. (1995), we developed a dynamic method based on statistical analysis of geophysical and biological data to delineate the boundaries of Longhurst's provinces in real time. This new method was applied to satelliteretrieved sea-surface temperature (SST) and phytoplankton biomass data averaged over the length of the research cruises (22, 17 and 16 d in spring, summer and fall respectively). CTD casts at various stations along the ship track were used to validate the method. A sensitivity study on primary production estimations performed with both definitions (static and dynamic) showed a negligible difference at the basin scale; however, variations at the province scale were significant. The regional description afforded by the new method was useful for the interpretation of expedition data.