Deposit size is a key factor in decisions concerning exploration and development of hydrocarbon reserves. Since it is unknown without an extensive study on individual field, there is a great interest in determining the field size distribution pattern, which could be used for reserves predictions. Geostatistics provides many models useful for such purposes (lognormal, exponential, Pareto, Zipf's law, etc.). One of the widely used models - lognormal distribution - was examined and its usefulness was confirmed in many cases. An alternative probability distribution for modelIing oil field size is the loghyperbolic, which could offer similar or even better fit to the empirical data. The aim of this paper is to examine both distribution models - lognormal and loghyperboIic - on the empirical sets of data. Three main Polish oil and gas basins (Carpathian, Forecarpathian and Poland Lowland) as well as three geological formations (Miocene, Dolomite, Rotliegende) were chosen to prove the qualities of both patterns. The lognormality and loghyperbolicity of observations were tested with x2 test of goodness of fit. The obtained results are compared and discussed in the paper. The results indicates that both models represent properly the distribution of oil and gas fields reserve size and could be recommend as useful geostatistical tools.