The role of the three principal epidemiological study designs-- descriptive, cohort and case-control studies--for the evaluation of risk, is illustrated by describing their use for the investigation of microbial keratitis. Descriptive studies have identified potential risk factors and causes of microbial keratitis by both case reports and case series; trauma, ocular surface diseases and, latterly, contact lenses have been identified as potential risk factors. These studies have also shown that Pseudomonas aeruginosa and Acanthamoeba are particularly associated with contact lens wear. However, these studies are limited because they cannot be used to quantify risk. Cohort studies, in which the number of cases of a disease developing in a defined and initially unaffected population are identified, are usually inappropriate for assessing rare conditions because the size of the study often has to be too large, and follow-up too long, to generate sufficient numbers of cases. Some of the disadvantages of this study type can be overcome by sampling techniques and have been successfully carried out to obtain incidence figures for microbial keratitis. Case-control studies use multivariable analyses to examine the risk of microbial keratitis associated with various factors. This is an economical study design for investigating rare diseases because a group of subjects with the disease is compared with a control group from the same population who are unaffected. The selection of an appropriate control group is a difficult problem in epidemiology but this study design has been crucial in identifying risk factors and potential causes of microbial keratitis.