The finite element approximation of the convection-diffusion-reaction equation suffers from numerical oscillations when diffusion is small and the Galerkin method is used. A family of methods can be designed to overcome this issue based on the Variational Multi-Scale (VMS) concept, which consists of splitting the unknown u as u = uh + u’, uh being the finite element solution and u’, which needs to be modelled, the sub-grid scale. We take u’ to be L2 orthogonal to uh, leading to the orthogonal sub-grid scale (OSGS) formulation [1]. A model for u’ is in fact a model for the error u - uh [2]. The analysis of VMS formulations is usually done in a mesh-dependent norm. We show that this norm with adequate scaling factors applied to u’ provides a robust a posteriori error estimator for the finite element solution. We provide an analysis proving that measuring the sub-grid scales yields a reliable error estimator. First, we adapt Verfürth’s approach [3] to our OSGS formulation and show rigorously that we can obtain an upper and a lower bound to the true error measured in a norm that contains information about the convective term. Next, we discuss several issues related to the analysis in the mesh-dependent norm associated to the formulation, highlighting the difficulties of this analysis. REFERENCES [1] R. Codina, S. Badia, J. Baiges and J. Principe, “Variational Multiscale Methods in Computational Fluid Dynamics”, in Encyclopedia of Computational Mechanics (eds E. Stein, R. Borst and T.J.R. Hughes), doi: 10.1002/9781119176817.ecm2117, John Wiley & Sons Ltd. (2017). [2] G. Hauke and D. Irisarri, “A review of VMS a posteriori error estimation with emphasis in fluid mechanics”, Computer Methods in Applied Mechanics and Engineering, 116341, (2023). [3] R. Verfürth, “Robust a posteriori error estimates for stationary convection-diffusion equations”, SIAM journal on Numerical Analysis, Vol. 43, pp. 1766−1782, (2005).