Barotropic variability of the Canadian Atlantic Shelf on synoptic time-scales is examined using data analysis, statistical and dynamical models. The main objective is to understand the connections between the subregions and develop models that can be used operationally. Data are used to determine dominant scales, run statistical models and validate dynamical models. Data assimilation is used to infer open boundary conditions for the dynamical models. In the first study synoptic variability of the Labrador Shelf is examined using a limited area model. Wind and upstream boundary forcing are found to be the most important sources of the variability. Based on model results a forecasting scheme is designed and tested. The second subregional study concentrated on the variability in the Gulf of Saint Lawrence. Boundary forcing via Cabot Strait is found to be the dominant source of variability. Data analysis indicates the presence of resonances in the Gulf, some of which can be related to wind and boundary forcing. In the last study, a large scale model of the Canadian Atlantic Shelf forced by wind and air pressure is used to examine connections between the regions. The model showed very good predictive skill for sea level and along-shore coastal currents. The wind driven flow from the Newfoundland Shelf is the most dominant source of remote variability for the Gulf of Saint Lawrence and the Scotian Shelf. High frequency signals from the Newfoundland Shelf can cross the Laurentian Channel and directly force the Scotian Shelf. Non-isostatic pressure forcing within the domain is negligible. The model driven by forecast winds can be run operationally to predict sea level and currents variability or supply boundary conditions for more sophisticated limited area model.