The road texture plays a fundamental role in the tyre/road noise generation. Bi-dimensional profiles are often used to evaluate the performances of road surfaces as regards the rolling noise emission. In order to predict noise levels from bi-dimensional longitudinal texture profiles, assumptions concerning the three-dimensional distribution properties of the road surface are required. If these assumptions are not met, a deviation may occur in the noise level evaluation. This paper aims at providing an insight into the expected radiated noise level variability due to possible texture anisotropy in the case of a slick tyre. The approach used in this paper is based on simulation tools intended to provide texture and noise data which are statistically analyzed afterwards. Three-dimensional gaussian random texture fields are numerically generated from a known bi-dimensional texture power spectral density assuming elliptical anisotropy. These texture fields are used as input of a dynamic rolling model that provides the radial vibration of the tyre belt. The resulting radiated noise levels are evaluated with a BEM model. The relationships obtained between the elliptical anisotropy rating and the calculated noise levels are presented and discussed.