Abstract In this work, we study the Galactic evolution of lithium by means of chemical evolution models in the light of the most recent spectroscopic data from Galactic stellar surveys. We consider detailed chemical evolution models for the Milky Way halo, discs, and bulge, and we compare our model predictions with the most recent spectroscopic data for these different Galactic components. In particular, we focus on the decrease of lithium at high metallicity observed by the AMBRE Project, the Gaia-ESO Survey, and other spectroscopic surveys, which still remains unexplained by theoretical models. We analyse the various lithium producers and confirm that novae are the main source of lithium in the Galaxy, in agreement with other previous studies. Moreover, we show that, by assuming that the fraction of binary systems giving rise to novae is lower at higher metallicity, we can suggest a novel explanation to the lithium decline at super-solar metallicities: the aforementioned assumption is based on independent constraints on the nova system birth rate, which have been recently proposed in the literature. As regards the thick disc, it is less lithium enhanced due to the shorter time-scale of formation and higher star formation efficiency with respect to the thin disc; therefore, we have a faster evolution and the ‘reverse knee’ in the A(Li) versus [Fe/H] relation is shifted towards higher metallicities. Finally, we present our predictions about lithium evolution in the Galactic bulge, which, however, still need further data to be confirmed or disproved.