A method based on space-filling coverage designs to optimize groundwater monitoring networks for plume detection and quantification is proposed. Space-filling objective functions are then compared with more classical functions. The method was applied to a hypothetical case-study with 160 candidate locations, resulting in final optimal design monitoring networks with 40 locations. Results show that the method is superior to those based strictly on the probability of contamination detection for quantifying maximum and mean values. In the light of these results fractal properties of space-filling coverage methods and of simulated annealing are also discussed.