The latest events such as the spread over Asia, Europe and Africa of the Highly Pathogenic Avian Influenza (HPAI) virus H5N1, the epidemic of A/H1N1pdm in 2009, the emergence of the Low Pathogenic Avian Influenza (LPAI) but zoonotic virus H7N9 in China in 2013 and the recent circulation of HPAI H5N8 in Europe, show that the permanent evolution of influenza virus in birds, humans and pigs is exposing the world to the risk of new strains with unpredictable consequences in public and animal health. In the last 10 years, a lot of efforts have been put in the improvement of capacity of animal and public health systems. However the disease is now endemic in several countries where the virus goes often undetected within the poultry population resulting in sporadic human cases and mortality. These countries are characterized by a large proportion of their population living in rural areas with poor incomes, a lack of primary care system and inefficient public or veterinary health sectors.Passive surveillance is often the only type of method feasible in poor rural settings in human and animal surveillance. With often incomplete, biased or delayed information this method will benefit from new methods of evaluation or new design concepts that could be implemented within a “One Health” framework to take into account the interfaces between human, animals and environment. We have in this thesis conceived and/or applied new methodologies for the evaluation, the design or the improvement of volunteer case-reporting of human or animal HPAI H5N1 in South-east Asia.We have first looked at different options of evaluation. We have applied stochastic scenario tree to model and assess the surveillance system of HPAI H5N1 in Thailand in backyard and free-range poultry production systems. We have estimated the sensitivity of the passive surveillance at 50% (CI95% 0.04-0.75) for a maximum detection of 3 infected farms, and showing the usefulness of this method to demonstrate freedom of disease in countries with limited resources. Thanks to participatory methods, we have involved Village Animal Health Workers (VAHWs) in their own evaluation and developed a new criteria grid, which includes local indicators of success developed and used by the VAHWs themselves.In a second part, we have considered methods to improve the design and the efficiency of passive surveillance. We have applied the grid conceived previously, to evaluate 251 VAHWs in three provinces of Cambodia. The grid allowed us to give a score to their level of activity and to analyse through logistic regression the factors influencing the most “good score”. Then we have implemented a pilot-study to test the use of SMS reporting from 112 participants from 68 villages, the objective was to detect peaks of mortality, to identify more rapidly outbreaks of infectious diseases. In a final section we have used multiple criteria decision analysis (MCDA) to map the risk of diffusion of HPAI H5N1 in poultry and in human, in order to adjust and reinforce the surveillance in the zones with greater risk of occurrence of the disease in Thailand and Cambodia.To conclude this work about tools and methods to improve surveillance systems in remote areas, we have done a comparative analysis of two challenging environments, Cambodia and Madagascar. We have done a cross analysis of the researches implemented by CIRAD (French Research Centre for International Development) in these countries during research projects implemented over the past decade. The objective was to show how new approaches for surveillance systems could be transferred between different countries with difficult socioeconomic environments and to propose new perspectives.