We discuss observational constrains coming from supernovae Ia [3] imposed on the behaviour of the Randall-Sundrum models. In the case of dust matter on the brane, the difference between the best-fit general relativistic model with a Λ-term [3] and the best-fit brane models becomes detectable for redshifts z > 0.6. It is interesting that brane models predict brighter galaxies for such redshifts which is in agreement with the measurement of the z = 1.7 supernova [6] and with the New Data from the High Z Supernovae Search Team [7]. We also demonstrate that the fit to supernovae data can also be obtained, if we admit the "super-negative" dark energy p = −(4/3)̺ on the brane, where the dark energy in a way mimics the influence of the cosmological constant. It also appears that the dark energy enlarges the age of the universe which is demanded in cosmology. Finally, we propose to check for dark radiation and brane tension by the application of the angular diameter of galaxies minimum value test. The idea of brane universes has originated from Hořava and Witten [1] followed by Randall and Sundrum [2]. Brane models admit new parameters which are not present in standard cosmology (brane tension λ and dark radiation U). From the astronomical observations of supernovae Ia [3, 4] one knows that the universe is now accelerating and the best-fit model is for the 4-dimensional cosmological constant density parameter Ω Λ (4) ,0 = 0.72 and for the dust density parameter Ω m,0 = 0.28 (index "0" refers to the present moment of time). In other words, only the exotic (negative pressure) matter in standard cosmology can lead to this global effect. On the other hand, in brane models the ̺ 2 quadratic contribution in the energy density even for a small negative pressure, contributes effectively as the positive pressure, and makes brane models less accelerating. In this paper we argue that in order to avoid this problem