I present a review of current ideas from colloid physics which might be relevant in order to understand the onset of crystallisation and amorphous aggregation processes in protein solution. In particular, the inadequacy of DLVO theory to account for all phenomenological aspects of crystallisation, such as salt-specificity, and for the basic features of the phase diagram, such as the presence of a metastable fluid–fluid separation, is discussed. The fundamental role of additional short-range attractive forces, microscopically orginating from the salting-out effect, is conversely stressed. In order to establish a simple model of protein interparticle interactions near crystallisation, I discuss some recent results obtained by our group for the osmotic compressibility of the metastable fluid phase of hen egg-white lysozyme. Light scattering measurements were performed in an extended volume fraction range at pH=4.7 as a function of temperature, adding NaC1 to screen the electrostatic interactions. The experimental compressibility up to particle volume fractions Φ≈0.23 is very successfully compared to the theoretical expression for a model of adhesive (“sticky”) hard spheres. This surprising quantitative agreement, obtained using a very simple form for the effective interparticle force, suggests that the thermodynamics of the system in the fluid phase is mainly determined by the very short-range nature of the interactions, and is rather insensitive to the detailed form of the potential.