We review the needs of the CoRoT /exoplanet program in terms of complementary observations. Such observations are indeed an essential element of the science of the CoRoT mission at different levels: prior to the launch for a characterization and a careful selection of the target stars in the exoplanet fields and during or after the CoRoT flight phase to affirm the planetary nature of detected transit events and to obtain planet and parent star properties in more detail. To prepare the sequence of CoRoT observations and to optimize the scientific analysis of the results, we built a stellar catalog of more than 10 millions of stars, based mainly on photometric observations from a large ground-based program we conducted for 3 years, and from existing catalogs. From optical and near-IR broad-band colors, we were able to perform the spectral classification of all stars within the range of V= 11 to 16. The method we used has the big advantage of allowing the observation of very large fields in a reasonable amount of telescope time and of providing reliable results on the spectral type and the luminosity class of about 80% of the stars. The quality of these estimates has been evaluated by comparison to the first results of a spectroscopic program carried out on a limited sample of stars. Our analysis not only allows to characterize the stellar population in the exoplanet fields but also provides, as an interesting by-product, new information to models of the Galactic stellar population close to the Galactic plane. All the preliminary information we collect on the target stars will be used during the CoRoT light curves analysis to remove ambiguities about the nature of the detected transiting object. In most cases, follow-up bservations are however still necessary to further filter out non-planetary transiting objects, as the large size of the CoRoT's PSF makes confusion with background eclipsing binaries likely. They are also mandatory to get independent verifications about the true nature of detected planets, as well as to obtain a maximum of information about them. This led us to set up a complete observing strategy, involving both radial velocity measurements and high spatial resolution photometry as well as possible space-based observations. This huge observational effort will also provide a unique and detailed basis for further statistical analysis of the properties of planets and their host stars, where preparatory observations constitute an unbiased and homogeneous sample, made with stars with and without planets.