Many features of biological populations can be described in terms of their heterogeneity by taking into account variations among individuals and cohorts in the population. In demography, the heterogeneity of populations can explain various features of age-dependent demographic observations including those related to mortality dynamics. Mortality dynamics is underlined by the Gompertz law stating that the mortality rate increases exponentially between sexual maturity and considerably old ages (i.e. between 20 and 80 years old). Deviations from the exponential increase are observed at early- and late-life intervals. Different models (i.e Heligman-Pollard model) were developed over the past decades to describe and explain these deviations. These models postulate that a few different processes take place in the population and affect its mortality dynamics. In this study we present a model based on an assumption that mortality dynamics is indeed underlined by the exponential law and the irregularities at young and very old ages are due to the heterogeneity of human population. We demonstrate that the model is capable of reproducing the entire pattern of mortality and explaining the deviations from the exponential growth. The model fitted to Swedish age-dependent mortality rates indicates that the population should be composed of four subpopulations each following the exponential law of mortality increase over age. We also expand the idea of heterogeneity to probability density and survival functions, that is we adjust the model to the number of Swedish deaths and survivors instead of mortality rates.