A rational design approach was taken to the planning and synthesis of a molecularly imprinted polymer capable of extracting caffeine (the template molecule) from a standard solution of caffeine and further from a food sample containing caffeine. Data from NMR titration experiments in conjunction with a molecular modelling approach was used in predicting the relative ratios of template to functional monomer and furthermore determined both the choice of solvent (porogen) and the amount used for the study. In addition the molecular modelling program yielded information regarding the thermodynamic stability of the pre-polymerisation complex. Post-polymerisation analysis of the polymer itself by analysis of the pore size distribution by BET yielded significant information regarding the nature of the size and distribution of the pores within the polymer matrix. Here is proposed a stepwise procedure for the development and testing of a molecularly imprinted polymer using a well-studied compound-caffeine as a model system. It is shown that both the physical characteristics of a molecularly imprinted polymer (MIP) and the analysis of the pre-polymerisation complex can yield vital information, which can predict how well a given MIP will perform. (c) 2006 Elsevier B.V. All rights reserved.