In the western Mediterranean Sea, the Northern Current is one of the major veins and is formed by the junction, in the Ligurian Sea, of the currents flowing northward along each side of Corsica; it is recognized as an entity along the continental slope as far as the Catalan Sea; surrounding the central zone where dense water formation occurs in winter. The seasonal and mesoscale variabilities of the Northern Current are analysed from a fortnightly hydrological survey carried out off Nice to a distance of similar to 55 km, from October 1990 to July 1991, and from similar to 30 current time series collected as deep as 2000 m, on four moorings set within a similar to 30 km coastal band from December 1990 to May 1991, in the framework of the PRIMO-O experiment. The hydrological surveys have mainly evidenced the spreading of Levantine Intermediate Water (LIW) in an on-offshore direction, the transformation of Modified Atlantic Water (MAW) into Winter Intermediate Water (WIW) in the deep winter, as well as the advection of less modified MAW. From a dynamic point of view, the seasonal variability is mainly depicted as a well-defined episode of narrowing, deepening and shoreward shift, from late January to mid-March, of a generally wide and shallow Northern Current. Currents have clearly appeared to be similar and highly correlated in an upper layer, the thickness of which is at least a hundred metres (between similar to 60 and 150 m), irrespective of the location of the points and of the season. This layer is expected to extend up to the surface. The flux of the Northern Current (between 0 and similar to 300 m) has ranged within 1-1.6 Sv from autum to summer, in agreement with the known values; relatively high values have been maintained during a rather long winter season. In accordance with other observations which have provided a description of a complete annual cycle, mesoscale activity is found to increase from autumn to deep winter and then displays a continuous decrease until summer at least. Mesoscale events have appeared to extend vertically over some few hundred metres, and the EOF decomposition has shown quasi no rotation of the fluctuations with depth. So, such events should have a relatively simple vertical structure, corresponding mainly to the first baroclinic mode with its zero-crossing at 400-500 m. Currents are relatively well represented by the barotropic and first baroclinic modes, the latter being predominant and more energetic, especially in winter. However, consequently to the variations of the Northern Current in width and depth, our most seaward point (similar to 30 km) is either outside or more or less within this current, so that the dynamic regime is generally more complex there. Nevertheless, the observed mesoscale events become, in deep winter, the barotropic ones of the central zone governed by vigorous convection. The fluctuations have generally shorter time scales in winter than in spring. Due to dramatic winter transformations, the Northern Current is mainly observed to be altered by instability processes, leading to meanders. Indeed, an important finding, based on the standard deviation ellipses and spectral analysis, is that its major fluctuations clearly appear to be quasi transverse within its own core. Moreover, from the polarization of energy, the signature of steep and large mesoscale meanders is clearly evidenced at 10-20 days. These meanders have far greater energy in winter than in spring, when they have slightly shorter periods (similar to 10 days). As previously observed, shorter fluctuations at 3-6 days with smaller amplitudes than those of the meanders at 10-20 days are also associated with a meandering current; they are expected to be intensified from spring-summer to the deep winter. In spring, while the flow is more stable, the predominant fluctuations resemble pulses of the current that may be expected to extend horizontally to a few tens of kilometres. Fundamental to coastal oceanographic problems is the fact that the circulation is actually unforeseeable in the coastal zone of similar to 10 km, where the main mesoscale phenomena have periods slightly shorter than in the core of the current. The major seasonal and mesoscale features of the Northern Current, especially the high flux values maintained during a relatively long winter season, i.e. during the period of dense water formation, its occasional narrowness and shoreward shift, and the extension of the central zone in deep winter, are probably very closely linked. Thus, as a hypothesis for further experimental and theoretical work, we suggest that the winter dense water formation should be effectively one of the major forcings of the circulation in the northern part of the western Mediterranean Sea.