In this article we review the mechanism of ocular surface staining. Water-soluble dyes are excluded from the normal epithelium by tight junctions, the plasma membranes and the surface glycocalyx. Shed cells can take up dye. A proportion of normal corneas show sparse, scattered time-dependent, punctate fluorescein uptake, which, we hypothesise, is due to a graded loss of the glycocalyx barrier, permitting transcellular entry into pre-shed cells. In pathological staining, there is little evidence of 'micropooling' at sites of shedding and the term `punctate erosion' may be a misnomer. It is more likely that the initial event involves transcellular dye entry and, in addition, diffusion across defective tight junctions. Different dye-staining characteristics probably reflect differences in molecular size and other physical properties of each dye, coupled with differences in visibility under the conditions of illumination used. This is most relevant to the rapid epithelial spread of fluorescein from sites of punctate staining, compared to the apparent confinement of dyes to staining cells with dyes such as lissamine green and rose bengal. We assume that fluorescein, with its lower molecular weight, spreads initially by a para-cellular route and then by transcellular diffusion. Solution-Induced Corneal Staining (SICS), related to the use of certain contact lens care solutions, may have a different basis, involving the non-pathological uptake of cationic preservatives, such as biguanides, into epithelial membranes and secondary binding of the fluorescein anion. It is transient and may not imply corneal toxicity. Understanding the mechanism of staining is relevant to the standardisation of grading, to monitoring disease and to the conduct of clinical trials. (C) 2014 Elsevier Ltd. All rights reserved.