A large body of research points to the biological importance of combinatorial post-translational modifications in proteins, such as the active role played by histone modification patterns in the development of cancers, neurodevelopmental disorders, neurodegenerative and other diseases. Nonetheless, our understanding of the precise biological function of different modification patterns is limited by the difficulty of identifying and quantifying different combinatorial isomers in their mixtures as they naturally occur. Tandem mass spectrometry, which infers primary structure from the mass-to-charge ratios of biomolecular fragments, is the preferred method of analysis for proteins and their post-translational modifications. However, the information contained in the mass-to-charge ratios of the individual fragments is frequently insufficient to identify the correct set of modification patterns present in a mixture of combinatorial isomers. This is because no possible single fragment of a combinatorially modified sequence is unique to that sequence in its mass-to-charge ratio. Here we show that the combinatorial post-translational modification problem can be solved by the recently introduced technique of two-dimensional partial covariance mass spectrometry, which provides information about fragment connectivity in a biomolecule by quantifying correlations between the random intensity fluctuations of its fragments, across repeated measurements. Unique fragment-fragment correlations provide the missing link between the non-unique individual fragments to produce unambiguous fingerprints of co-occurring combinatorial isomers, enabling the discovery of biomolecular combinatorial modification patterns by mass spectrometry.