To make the Coffee cDNA-Macroarrays, a non-reduntant group of clones (Unigene) was initially generated by bioinformatics analysis of the Brazilian Coffee Genome Database. The re-arraying routine was performed with the Q-Bot system (Genetix) and resulted in a non-redundant group of 33,000 clones organized in 86 plates (384). Plasmid DNA from all 33 thousand clones was extracted and representative samples were analyzed by gel electrophoresis. To check the accuracy of the re-arraying routine, 4 samples per plate were sequenced and compared by BlastN analysis against the Coffee Database. The results of this analysis indicated 8% mistakes during the re-arraying routine. Plates containing mistakes were re-arrayed and novel samples sequenced. The next step for the Macroarray construction is the printing of the membranes and the objective of this work was to establish the best protocol for this routine. The type of DNA (plasmid vs. PCR product) and the concentration were the variables evaluated. The results obtained with the hybridization experiments indicated that in general the printing routine with the Q-Bot system was very uniform and accurate among the replicates. In addition, the data obtained indicated that the radioactive signals obtained with the PCR product-spots were stronger than the ones obtained with plasmid DNA due to the different molar ratio and probably also, due to the better denaturing step of a linear DNA fragment.