The typical treatment of the small-angle X-ray scattering (SAXS) experiment for supercritical fluids contains two interconnected uncertainties. One supposes a priori an unimodal, Gaussian-type distribution of local density for any volumes of observation in spite of its strong fluctuations near and above the actual critical point. Accordingly, the empirical or semitheoretical unified equations of state (EOS) are used to normalize the SAXS-intensities which cannot express the singular behavior of the isothermal compressibility and, as a result, cannot predict adequately the Ornstein-Zernike (OZ) correlation length. On the other hand, the numerical molecular-dynamical (MD) experiment in a combination with the coarse-graining analysis reveals the typically bimodal structure of a compressible fluid regime in the supercritical region if one takes into account not only the aggregates of particles with the augmented average density but also the steady existence of "voids" with its depletion. To reconcile qualitatively laboratory SAXS- and numerical MD-experiments the methodology of global fluid asymmetry (GFA) proposed by authors in the framework of FT-(fluctuational thermodynamics) model has been used in this work. In particular, the known problem of a "ridge" formed by the supposedly Gaussian distributions of the isothermal density fluctuations is reformulated in terms of their presumably bimodal structure. Its location below the critical isochore in the P,T-diagram found by the traditional SAXS-treatment is independently confirmed only for the liquid like peak of supercritical distribution while the additional, predicted here gas like peak is definitely located at the lower densities. The both peaks of such thin structure revealed in the density fluctuations coincide at the pseudocritical (but non-mean-field) point where the distributions of any thermodynamic densities become indeed Gaussian ones at the higher temperatures and pressures of a ridge. The phenomenological generalization of the OZ-theory based on the FT/GFA-approach is formulated in terms of the finite-range direct and total correlation integrals. (C) 2014 Elsevier B.V. All rights reserved.