The observations of Li abundances in pre-Main-Sequence, Main-Sequence (Population I and II), sub-giant, and giant stars are reviewed in order to show how Li can be used as a constraint on stellar hydrodynamics and in particular on particle transport processes in stars. Important observational results include the tight Li abundance dependence on T eff in the Hyades, the time dependence of the Li abundance below T eff = 6000 K, the presence of a Li gap at T eff = 6700 K in young clusters and the large Li abundance in some peculiar giants. The observed abundances are compared to models which include progressively more physical processes. The ‘standard’ stellar evolution model is compatible with the upper envelope of the observations in young clusters such as the Pleiades and α Per. The observed Li underabundances is then caused by Li burning on the pre-Main Sequence. The large abundance spread observed is not understood. It does not appear to be simply related to rotation since the Pleiades stars rotate more slowly but have larger Li abundances than many stars of α Per. The Li abundance gap observed in clusters is not explained by the ‘standard’ model. Models involving diffusion seem to explain it in a natural way, though meridional circulation could also be involved. Evolutionary effects and the interaction between diffusion and meridional circulation should, however, be taken more fully into account in those models. The Li abundances in giants show that additional destruction processes are involved beyond those included in the ‘standard’ evolutionary models. Meridional circulation is compatible with most of those observations, without any arbitrary parameter being adjusted. While turbulence is nearly certainly present in stars, it is poorly understood and we suggest that it should be invoked to explain only those phenomena that the better understood processes cannot explain. Its description always involves arbitrary parameters. Turbulence appears to be required to explain the Li abundances in the Sun and in G stars of the Hyades and older clusters. In halo stars, the observed Li abundance has probably been reduced from the original by a factor of 2 so that the original abundance was probably equal to log N(Li) = 2.5. More calculations are needed to better establish this value. The large Li abundances observed in some peculiar giants are not understood.