his paper deals with online scheduling of a flexible job-shop with blocking where jobs are transported between the machines by one or several automated guided vehicles. The objective is to minimise the total weighted tardiness of the jobs. Schedules have to be generated online (i.e., jobs only become known to the scheduler at their respective release dates) as well as in real time (i.e., only a limited amount of computation time is allowed). We propose a MIP formulation for the offline situation and an online Tabu search algorithm that decomposes the problem into an assignment and a scheduling subproblem. In some computational experiments online and offline algorithms are compared for various problem instances.