Due to the success of DNA microarrays and the growing numbers of available protein expression clones, protein microarrays become more and more popular for the high-throughput screening of protein interactions. However, the widespread applicability of protein microarrays is currently hampered by the large effort associated with their production. Apart from the requirement for a protein expression library, expression and purification of the proteins themselves and the lacking stability of many proteins remains the bottleneck. Here, we present an approach that allows the generation of high-density protein microarrays from unbound DNA template molecules on the chip. It is based on the multiple spotting technique (MIST) and comprises the deposition of a DNA template in a first spotting step and the transfer of a cell-free transcription and translation mix on top of the same spot in a second spotting step. Using wt-GFP as a model protein, we demonstrate the time and template dependence of this coupled transcription and translation and show that enough protein is produced to yield signals that are comparable to 300 µg/ml of spotted protein. Plasmids as well as unpurified PCR products can be used as templates and as little as 35 fg of PCR product (~ 22500 molecules) are sufficient for the detectable expression of full-length wt-GFP in subnanoliter volumes. We show that both, APTES and Ni-chelate surfaces can be used for capture of the newly synthesized proteins. Surprisingly, we observed that Ni-chelate coated slides are binding the newly synthesized proteins in an unspecific manner. Finally, we adapted the system to the high-throughput expression of libraries by designing a single primer pair for the introduction of the required T7 promoter and demonstrate the in situ expression using 384 randomly chosen clones.