Recent marine and geophysical research has demonstrated that long-term time series of critical parameters are needed to understand the various ocean systems. ASSEM is the first application of a new concept of sea bed observatories dedicated to long term monitoring of a small area (a few km/sup 2/), lying on a network of interconnected measurement nodes. It is a project that enhances marine technologies allowing real time monitoring of the sea bed. The main component of each node is the Communication and STOrage Front end, COSTOF providing a set of enhanced sensors (pore pressure, methane, geodesy, tilt, CTD, turbidity, currents, ...) with the means to communicate with the external world through an underwater acoustic or cabled network, and to locally store the produced data. Alarms can also be generated by processing these data. The architecture is organized around an internal CAN/CANopen bus, hosting sensors, communication and memory devices on a common transmission backbone. The software resources enabling a monitoring node to act as a network node (routing algorithms throughout the network, network configuration management, data transmission protocol and other network layers) are implemented in every COSTOF. Alarms can be generated for example if a critical parameter, or a group of parameters, exceeds a programmed threshold for a given time. The evaluation of this new concept will be made through two experiments addressing two sea bed problems: slope instability risks and seismic risks. The first one will take place at a site in Norway offering a risk slope instability. The experiment should include hook-up of integrated pore pressure sensors, installed in bore-holes, and set-up of standard sensor interface and communication system for data transfer to the surface. The second experiment will take place in the Gulf of Corinth. The shelf, slope and margin of the basin off the coast of a faulted area is selected for the deployment of the ASSEM array of sensors. It i s the most active extensional basin in Europe, with high rates of margin uplift This concept could be applied to other domains such as long-term studies in biology or emergency monitoring.