Proteome profiling techniques rely on the separation of proteins or peptides and their subsequent quantification. The reliability of this technique is still limited because a proteome profiling result does not necessarily represent the true protein composition of the analysed sample, thus seriously hampering proper data interpretation. Many experimentally observed proteome alterations are biologically not significant. It was the aim of this study to use the knowledge of the biological context of proteins in order to establish optimised proteome profiling protocols. While 2-D spot patterns of total cell protein fractions were found to poorly represent the true protein composition, purified subcellular protein fractions were found to better represent the protein composition of the analysed sample. The application of a standardised protocol to different kinds of cells revealed several striking observations. Firstly, the protein composition of cultured cells of various origins is very similar. Secondly, proteome alterations observed with the described protocols do make sense from a biologic point of view and may thus be considered as truly representative for the analysed samples. Thirdly, primary white blood cells isolated from different donors were found to show minor, but reproducible and significant individual differences. We designate the consideration of known properties of identified proteins in proteome profiles as a knowledge-based approach. The present data suggest that this approach may tremendously help to improve the applied techniques and assess the results. We demonstrate that the fulfilment of well-defined criteria of proteome profiles eventually results in reliable and biologically relevant data.