In this paper, a new approach for leak detection in water networks is proposed which considers an extended time horizon analysis of pressure sensitivities. Previous works based on pressure sensitivities analysis were developed by considering time instant evaluation. This fact makes them very sensitive to demand changes and noise in measurements. The proposed approach has been combined with five detection methods: the first is the binarization method, the second, third and fourth are based on the analysis of vectors with methods of correlation, vector angle and Euclidean distance respectively, and finally, the fifth is based on the least square optimization method. Another contribution of this paper is the performance comparison between the five detection methods in presence of single leaks in scenarios with noise in measurements and demands in each node. Results showed that in most of the methods efficiency is high, being the best the vector angle method, with efficiencies higher than 96%. The correlation and optimization method had similar behaviors with efficiencies superior to 90%. Finally, the binarization method is effective only in some scenarios but in presence of noise has a poor performance.