This chapter summarizes the different approaches used to reconstruct past environmental conditions from micropaleontological proxy data. The first group of methods involves calibration, i.e. a relationship is calculated between a given environmental variable and the abundances of the various species (e. g. regression, partial least squares). The second group involves the calculation of a similarity index between the assemblages (modern analogue technique on raw assemblages or based on surface responses). The validation of the transfer functions must be done using several well known statistics calculated on independent modern data (coefficient of determination, root-mean-square error.). In conclusion, limitations and potentials of the different approaches are discussed.