We describe a strategy, which we term hypothesis-driven multiple-stage mass spectrometry (HMS-MS), for the sensitive detection and identification of phosphopeptides derived from enzymatic digests of phosphoproteins. In this strategy, we postulate that any or all of the potential sites of phosphorylation in a given protein may be phosphorylated. Using this assumption, we calculate the m/z values of all the corresponding singly charged phosphopeptide ions that could, in theory, be produced by the enzyme employed for proteolysis. We test ions at these m/z values for the presence of phosphoserine or phosphothreonine residues using tandem mass spectrometry (MS(2)) in a vacuum MALDI ion trap mass spectrometer, where the neutral loss of the elements of H(3)PO(4) (98 Da) provides a sensitive assay for the presence of phosphopeptides. Subsequent MS(3) analysis of the (M + H - 98)(+) peaks allows us to confirm or reject the hypotheses that the putative phosphopeptides are present in the sample. HMS-MS was successfully applied to the detection and identification of phosphopeptides from substrates of the Saccharomyces cerevisiae cyclin-dependent kinase (Cdk) Cdc28, phosphorylated in vitro (Ipl1) and in vivo (Orc6), basing hypothesis formation on the minimal Cdk consensus phosphorylation motif Ser/Thr-Pro. The method was also used to find in vitro phosphopeptides from a domain of the Drosophila melanogaster protein PERIOD, hypothesizing possible phosphorylations of all Ser/Thr residues without assuming a consensus motif. Our results demonstrate that HMS-MS is a sensitive, highly specific tool for systematically surveying proteins for Ser/Thr phosphorylation, and represents a significant step toward our goal of comprehensive phosphorylation mapping.