This paper deals with classification of dose distributions of nuclear workers based on antikurtosis (Q) and entropy coefficients (K) and their relationship presented in QK-diagrams. It is shown that determination of the most appropriate distribution to adopt, for a specific data set of a wide range of input data, requires building and analysing QK-diagrams for distributions of logarithms of individual doses. Actual dose distributions for emergency and occupational exposure situations were then considered, as well as doses for one day of work during clean-up and routine activities. It is shown that, in all cases, three types of distributions of logarithms of individual doses were present: normal, Weibull and Chapeau. The location of the representation point of a dose distribution reflects the degree of dose control of the group of workers whose individual doses are collectively displayed in the QK-diagram. The more the representation point of the analysed distribution of the logarithms of the individual dose of a given contingent of workers deviates from the point of the lognormal distribution, the more there was intervention in the process of individual dose accumulation. Thus, QK-diagrams could be used to develop a dose control function. It is shown that the hybrid lognormal distribution, which is widely used in the field of radiation safety, for the purpose of approximation of real dose distributions, is unable to satisfactorily describe many dose distributions arising in aftermath operations and occupational exposure.