The influenza A H1N1 epidemic has spread rapidly worldwide on account of the current conditions of high interconnectivity and transport speed both among people and countries. The spatial spread of the epidemics can be explained by the percolation theory which allows to estimate a threshold beyond which the transmission of the infection among different geographic regions occurs. The aim of this study was to test the predictive ability of the percolation model of influenza A H1N1 epidemic in Argentina according to data gathered by the National Department of Public Health. In the model, the country was considered as a set of irregular, contiguous and continuous geometric figures, which can be represented in two dimensions on a plane. We analyzed the proportion of infected provinces at the moment of percolation in relation to time in days and compared observed and expected values by curvilinear equations in a logistic model. Percolation occurred on day 45. The expected value generated by the model was 42.4 days, 95 % CI 28.5 to 56.3. The difference between observed and expected values was p = 0.997. We conclude that the model has good fit and predictive capacity.