We study the usefulness of lookahead in online server routing problems: if an online algorithm is not only informed about the requests released so far, but also has a limited ability to foresee future requests, what is the improvement that can be achieved in terms of the competitive ratio? We consider several online server routing problems in this setting, such as the online traveling salesman and the online traveling repairman problem. We show that the influence of lookahead can change considerably depending on the particular objective function and metric space considered. Key words: on-line algorithms, competitive analysis, lookahead, traveling salesman problem