Biomarker discovery in clinical proteomics is being performed on relatively large patient cohorts by utilizing the high throughput of laser desorption/ionization mass spectrometry (MALDI- and SELDI-TOF-MS). Dealing directly with patient samples as opposed to working in cell or animal systems requires a host of considerations both before and after mass spectrometric analysis to obtain robust biomarker candidates. The challenges associated with the heterogeneity of typical samples are amplified by the ability to detect hundreds to thousands of proteins simultaneously. Adherence to protocols and consistency, however, can ensure optimal results. A study starts necessarily with a relevant clinical question and proceeds to a planning phase where sample availability, statistical test selection, logistics and bias reduction are key points. The physical analysis requires consistency and standardized protocols that are helped significantly through automation. Data analysis is broken into two stages, screening and final testing, which can detect either single candidates or a pattern of proteins. Biomarker identification can be performed at this point and will help significantly in the last stage, interpretation. Replication should be performed in an independent sample set in a separate study. The candidate biomarkers from an initial study give a wealth of information that can help to pinpoint patient subpopulations for a more exhaustive proteomic study using complementary platforms with limited capacity but extremely high information content. A clinical proteomics pilot project can also lead to better selection of model systems by providing a direct link with patient samples.