Protein stable isotope fingerprinting (P-SIF) is a method to measure the carbon isotope ratios of whole proteins separated from complex mixtures, including cultures and environmental samples. The goal of P-SIF is to expose the links between taxonomic identity and metabolic function in microbial ecosystems. To accomplish this, two dimensions of chromatography are used in sequence to resolve a sample containing ca. 5-10 mg of mixed proteins into 960 fractions. Each fraction then is split in two aliquots: The first is digested with trypsin for peptide sequencing, while the second has its ratio of (13)C/(12)C (value of δ(13)C) measured in triplicate using an isotope-ratio mass spectrometer interfaced with a spooling wire microcombustion device. Data from cultured species show that bacteria have a narrow distribution of protein δ(13)C values within individual taxa (±0.7-1.2‰, 1σ). This is moderately larger than the mean precision of the triplicate isotope measurements (±0.5‰, 1σ) and may reflect heterogeneous distribution of (13)C among the amino acids. When cells from different species are mixed together prior to protein extraction and separation, the results can predict accurately (to within ±1σ) the δ(13)C values of the original taxa. The number of data points required for this endmember prediction is ≥20/taxon, yielding a theoretical resolution of ca. 10 taxonomic units/sample. Such resolution should be useful to determine the overall trophic breadth of mixed microbial ecosystems. Although we utilize P-SIF to measure natural isotope ratios, it also could be combined with experiments that incorporate stable isotope labeling.