This report describes the development of a system capable of managing jobs in NorduGrid. The project is a continuation of a former project, in which a daemon capable of resubmitting jobs were developed. The report starts by giving an introduction to grids, what they are, and how resource sharing is organized by creating virtual organizations. Hereafter a general grid model is presented, along with examples of three different grid architectures. The NorduGrid project is then described with regards to its political structure and architecture of the toolkit. After the description of the NorduGrid project, our initial considerations for the project is described. The need for a job manager is analyzed, our design philosophy is presented and features of the Job Manager discussed. Hereafter an overview of the Job Manager is presented, where after the small modules in the Job Manager are discussed. The next chapter describes how Job Management is done by separating bookkeeping of jobs and actions into different modules, making it possible to redefine the way jobs are dealt with. It is then described how to avoid the Job Manager becoming a single point of failure, by making other Job Managers acting as ailover. Lastly a future work chapter presents further ideas. Finally we conclude that the Job Manager delivers a framework for production systems while also providing applications with a simple interface to access the grid.