Free healthcare obviously works when a partner from abroad supplies a health centre or a health district with medicines and funding on a regular basis, provides medical, administrative and managerial training, and gives incentive bonuses and daily subsistence allowances to staff. The experiments by three international NGO in Burkina Faso, Mali and Niger have all been success stories. But withdrawing NGO support means that health centres that have enjoyed a time of plenty under NGO management will return to the fold of health centres run by the state in its present condition and the health system in its present condition, with the everyday consequences of late reimbursements and stock shortages. The local support given by international NGOs has more often than not an effect of triggering an addiction to aid instead of inducing local sustainability without infusion. In the same way, scaling up to the entire country a local pilot experiment conducted under an NGO involves its insertion into a national bureaucratic machine with its multiple levels, all of which are potential bottlenecks. Only experiments carried out under the "ordinary" management of the state are capable of laying bare the problems associated with this process. Without reformers 'on the inside' (within the health system itself and among health workers), no real reform of the health system induced by reformers 'from the outside' can succeed.