SUMMARYWhen the abnormality of a skin response to sunlight is expressed in terms of a decreased minimum erythemal dose (MED) on photo‐testing, it is essential to know the range of response in normal skin. In any homogeneous population, the statistical distribution of the MED is skewed and is not Gaussian (normal), but a Gaussian distribution tan be fitted to the logarithm of the MED. Using established methods of probit analysis, such ‘lognormal’ distributions have been fitted to the observed proportions of erythemal responses to a series of test exposure doses of UV radiation (250–365 nm), thus providing estimates of the average log MED for normal skin, the standard deviations of the distributions and associated confidence limits. These estimates, in arithmetical terms, are available for quantitative comparison with the results of other test exposures on skin which may be a