The quantitative attributes of human leukocyte proteins detected by silver staining two-dimensional electrophoresis (2-DE) gels were studied by using computer-assisted data analysis. Experiments included (a) analysis of replicate patterns of the same sample, (b) analysis of different dilutions of the same sample, and (c) analysis of samples from different individuals. Over 200 proteins were observed to have coefficients of variation (CV) less than or equal to 15% when data from replicate patterns were analyzed. In contrast, 8 proteins had CV values of less than or equal to 15% when data from different samples were analyzed. The dilution experiment showed that a majority of the proteins detected with some consistency (i.e., observed in at least 80% of the patterns) have a linear relationship between the amount of protein loaded onto a 2-DE gel and the spot volume in the final 2-DE pattern. The slope of the curves and the deviation from linearity were found to be quite protein-specific. These results indicate that optimization of sample purity and minimization of staining protocol variables are required to limit the background quantitative variability between and within 2-DE runs to a level that will allow detection of quantitative changes indicative of biological responses.