This thesis is about challenges in Information Systems (IS) development in developing countries contexts, with an emphasis on challenges faced in processes dealing with improvement and replacement of Health Information Systems. Empirical data are collected from action research interventions in the Mozambican health sector. The work was done under the umbrella of a global action research project, called the Health Information System Programme. I argue that typical priorities among stakeholders in centralized developing countriessettings, as focus on immediate needs rather than long term solutions, control, and reluctance in support for significant changes, may lead to the unintended consequences of creation and maintenance of legacy systems. Lack of communication and consideration of real users needs among decision makers is a contributor to this, meaning that centralized contexts may be especially challenged in HIS projects, as the distance between decision makers and real users typically is substantial organizationally-wise. The research suggests that the dilemma of introducing small changes with a new IS and at the same time supporting further organizational development may be solved by implementing flexibility in the IS done practically through flexibility in design and by enabling implementation of emerging needs by looking upon systems development as a continuous process through the life time of the IS. By applying different change strategies for different levels of the installed base, getting away from legacy systems problems may be reached in change resisting social systems.