Up to now, many approaches for estimating the statistics of road traffic noise have been carried out by the introduction of several models, e.g., an equally spaced vehicles model, an exponentially distributed vehicle model, and an Erlang distribution type model in a simplified sound propagation environment such as a free sound field. However, in several studies based particularly on the latest Erlang distribution type model (i.e., gamma headway distribution model), only the first‐ and second‐order moments of the sound intensity fluctuation which can be derived from the statistical information on the location of merely one and/or two vehicles (e.g., the distributions for the position of the vehicle and the distance between two arbitrary vehicles flowing in the same direction) are taken individually into consideration. On the other hand, the higher‐order statistical properties of traffic noise are rather important in order to investigate the whole shape of the noise distribution form from which any noise evaluation index can be derived. Thus, in this paper, our main interest is devoted to considering quantitatively the relation between the multidimensional correlation properties of the sound intensity and the higher‐order information on the vehicles flow by use of Stratonovich's stochastic theory for a random points system. Furthermore, the relation between our theoretical result and well‐known previous studies is discussed with experimental confirmation for several lower‐order moments.