The ozone layer has been under surveillance since the discovery of the ozone hole over Antarctica in the early eighties. Since then, the abundance of ozone-depleting substances (ODS) regulated by the Montreal Protocol have peaked by the mid or end of the nineties depending on latitudes and have now started a slow decrease considering the atmospheric lifetimes of these compounds. As a consequence, the ozone layer has stabilized and shows early signs of recovery. While the decrease of ODS content is expected to be the dominant cause of the future long-term evolution of stratospheric ozone, climate change effect and natural variability of the atmosphere still hinder the unambiguous detection of recovery expected from ODS decrease. Indeed, the very low ozone content observed during the Arctic winter 2010/2011 was caused by a very peculiar dynamical situation in the Northern Hemisphere, together with the still high levels of ODS in the stratosphere (WMO, 2014). The detection of the small signal of recovery is also a challenge for the current ozone global observing system, in particular regarding ozone vertical distribution. In that case, quantification of long-term ozone trends requires the combination of satellite records characterized by different frequency sampling and vertical resolution, after careful verification of their long-term stability and mutual consistency using reference ground-based monitoring networks. This presentation will review the current understanding of ozone recovery in various stratospheric regions, both regarding its total content and vertical distribution.