Abstract Photosensitisers are the photoactive molecules used in photodynamic therapy (PDT) of cancer. Despite the importance of their interaction with polypeptides, only the binding to plasma proteins has been investigated in some detail. In our study we compared the binding of Protoporphyrin IX (a clinically useful photosensitiser) to an immunoglobulin G, with the binding to albumins. Binding to IgG is relevant because a possible method of increasing tumour specificity of photosensitisers is to bind them to tumour-specific antibodies. Binding constants to albumins and the immunoglobulin were comparable (≅6×10 −6 M −1). The apparent number of PPIX molecules bound to each protein was also within a similar range (from 4 to 7). The absence of a shift in the emission spectrum of PPIX bound to IgG, however, indicates that either larger aggregates of PPIX bind to the immunoglobulin or that the binding site leaves PPIX exposed to the buffer. We observed that PPIX photoproducts compete with PPIX for the same binding sites. The number of PPIX molecules bound to each protein in the presence of photoproducts decreased by 50–80%. Due to the spectral overlap between PPIX and its photoproducts, the binding in the presence of photoproducts was investigated using Derivative Synchronous Fluorescence Spectroscopy (DSFS) to improve the spectral separation between chromophores in solution. We also concluded that fluorescence measurements underestimate the number of PPIX molecules binding each protein. In fact, non-linear Scatchard plots (in the case of albumin binding) by definition yield a minimum number of molecules attached to a protein. Moreover, the binding of large aggregates, formed by an unknown number of PPIX molecules, to IgG results in the underestimate of the number of molecules bound. The number of PPIX molecules bound to these proteins is also much larger than the number of sites estimated by protein fluorescence quenching.