The authors designed a risk-based approach to the selection of poultry flocks to be sampled in order to further improve the sensitivity of avian influenza (AI) active surveillance program in Cuba. The study focused on the western region of Cuba, which harbours nearly 70% of national poultry holdings and comprise several wetlands where migratory waterfowl settle (migratory waterfowl settlements - MWS). The model took into account the potential risk of commercial poultry farms in western Cuba contracting from migratory waterfowl of the orders Anseriformes and Charadriiformes through dispersion for pasturing of migratory birds around the MWS. We computed spatial risk index by geographical analysis with Python scripts in ESRI® ArcGIS 10 on data projected in the reference system NAD 1927–UTM. Farms located closer to MWS had the highest values for the risk indicator pj. and in total 31 farms were chosen for targeted surveillance during the risk period. The authors proposed to start active surveillance in the study area 3 weeks after the onset of Anseriformes migration, with additional, sampling repeated twice in the same selected poultry farms at 15 days interval ( Comin et al., 2012 and EFSA, 2008) to cover the whole migration season. In this way, the antibody detectability would be favoured in case of either a posterior AI introduction or enhancement of a previous seroprevalence under the sensitivity level. The model identified the areas with higher risk for AIV introduction from MW, aiming at selecting poultry premises for the application of risk-based surveillance. Given the infrequency of HPAI introduction into domestic poultry populations and the relative paucity of occurrences of LPAI epidemics, the evaluation of the effectiveness of this approach would require its application for several migration seasons to allow the collection of sufficient reliable data.