Abstract : This thesis deals with intermittent target search strategies, which combine slow phases, allowing the searcher to detect the target, and fast phases without detection. Foraging animals are an example at the macroscopic scale. We propose a model, alternative to the famous Lévy strategies, and show analytically that the mean search time can be minimized as a function of the mean duration of both phases. Our first example at the microscopic scale is given by proteins searching for targets on DNA. We analytically calculate the distribution of the distance travelled along DNA during a 3D excursion, adapt it to a single-molecule experiment and show that the observed trajectories combine 1D and 3D diffusion. Another cellular example is provided by active transport of vesicles, which diffuse or bind to motors performing ballistic motion.We optimize the global kinetic constant within a general framework of reactions limited by this kind of transport. Finally, these intermittent strategies could constitute a generic search mechanism. We systematically study the influence both of the modeling of the detection phase and of the space dimension, and show that the optimality of intermittent strategies is a robust result.