We report the first metabolic labeling of Arabidopsis thaliana for proteomic investigation, demonstrating efficient and complete labeling of intact plants. Using a reversed-database strategy, we evaluate the performance of the MASCOT search engine in the analysis of combined natural abundance and 15N-labeled samples. We find that 15N-metabolic labeling appears to increase the ambiguity associated with peptide identifications due in part to changes in the number of isobaric amino acids when the isotopic label is introduced. This is reflected by changes in the distributions of false positive identifications with respect to MASCOT score. However, by determining the nitrogen count from each pair of labeled and unlabeled peptides we may improve our confidence in both heavy and light identifications.