The recent years have seen tremendous improvement of our understanding of high resolution reachable in TERS experiments, forcing us to re-evaluate our understanding of the intrinsic limits of this field, but also exposing several inconsistencies. On the one hand, more and more recent experimental results have provided us with clear indications of spatial resolutions down to a few nanometres or even subnanometre scale. Moreover, lessons learned from the recent theoretical investigations clearly support such high resolutions, and vice versa the obvious theoretical impossibility to evade high resolution from a purely plasmonic point of view. On the other hand, most of the published TERS results still, to date, claim a resolution on the order of the tens of nanometres that would be somehow limited by the tip apex, a statement well accepted for the past 2 decades. Overall, this now leads the field to a fundamental question: How can this divergence be justified? The answer to this question brings up an equally critical one: How can this gap be bridged? This review aims at raising a fundamental discussion related to the resolution limits of tip-enhanced Raman spectroscopy, at revisiting our comprehension of the factors limiting it both from a theoretical and experimental point of view and at providing indications on how to move the field ahead. It is our belief that a much deeper understanding of the real accessible lateral resolution in TERS and the practical factors that limits them will simultaneously help us to fully explore the potential of this technique for studying nanoscale features in organic, inorganic and biological systems, but also to improve both the reproducibility and the accuracy of routine TERS studies. A significant improvement of our comprehension of the accessible resolution in TERS is thus critical for a broad audience, even in certain contexts where high resolution TERS is not the desired outcome.