While collisionally activated dissociation (CAD) pathways for peptides are well characterized, those of intact proteins are not. We systematically assigned CAD product ions of ubiquitin, myoglobin, and bovine serum albumin generated using high-yield, in-source fragmentation. Assignment of Ͼ98% of hundreds of product ions implies that the fragmentation pathways described are representative of the major pathways. Protein dissociation mechanisms were found to be modulated by both source declustering potential and precursor ion charge state. Like peptides, higher charge states of proteins fragmented at lower energies next to Pro, via mobile protons, while lower charge states fragmented at higher energies after Asp and Glu, via localized protons. Unlike peptides, however, predominant fragmentation channels of proteins occurred at intermediate charge states via non-canonical mechanisms and produced extensive internal fragmentation. The non-canonical mechanisms include prominent cleavages Cterminal to Pro and Asn, and N-terminal to Ile, Leu, and Ser; these cleavages, along with internal fragments, led to a 45% increase in sequence coverage, improving the specificity of top-down protein identification. Three applications take advantage of the different mechanisms of protein fragmentation. First, modulation of declustering potential selectively fragments different charge states, allowing the source region to be used as the first stage of a low-resolution tandem mass spectrometer, facilitating pseudo-MS 3 of product ions with known parent charge states. Second, development and integration of automated modulation of ion funnel declustering potential allows users access to a particular fragmentation mechanism, yielding facile cleavage on a liquid chromatography timescale. There are two fundamental differences between the so-called top-down and the traditional bottom-up approaches (bottom-up uses proteolytic peptides of Յ3000 Da) [3]. The first is that top-down methods can provide the mass of the protein and the second is that top-down proteomics methods rely exclusively upon fragments generated in the gas phase, whereas bottom-up methods employ an additional chemical or enzymatic digestion step [3]. Limitations of top-down methods are the upper mass limit (currently at 669 kDa [4] for protein identification) and the efficiency, and understanding of the mechanism of, gas-phase protein fragmentation.Numerous surface-induced dissociation and collisionally activated dissociation (CAD) studies [5][6][7][8][9][10][11][12] characterized peptide fragmentation mechanisms [5], including the comprehensive characterization of fragmentation trends by systematic, large-scale studies [13][14][15]. From these studies arose the "mobile proton model" (and the recent "pathways in competition" model), which state that fragmentation requires the collisionally activated transfer of a proton from a basic site to a less basic site. The positions of amino acids with high structural rigidity and high gas-phase basicity (proline, arginine, histidine,...