This paper reports on the recent studies of toroidal and poloidal momentum transport in JET. The ratio of the global energy confinement time to the momentum confinement is found to be close to tau(E)/tau(phi) = 1 except for the low density or low collisionality discharges where the ratio is tau(E)/tau(phi) = 2-3. On the other hand, local transport analysis of around 40 discharges shows that the ratio of the local effective momentum diffusivity to the ion heat diffusivity is chi(phi)/chi(i) approximate to 0.1-0.4 (averaged over the radial region r/a = 0.4-0.7) rather than unity, as expected from the global confinement times and used often in ITER predictions. The apparent discrepancy in the global and local momentum versus ion heat transport can be at least partly explained by the fact that momentum confinement within edge pedestal is worse than that of the ion heat and thus, momentum pedestal is weaker than that of ion temperature. In addition, while the ion temperature profile shows clearly strong profile stiffness, the toroidal velocity profile does not exhibit stiffness, as exemplified here during a giant ELM crash. Predictive transport simulations with the self-consistent modelling of toroidal velocity using the Weiland model and GLF23 also confirm that the ratio chi(phi)/chi(i) approximate to 0.4 reproduces the core toroidal velocity profiles well and similar accuracy with the ion temperature profiles. Concerning poloidal velocities on JET, the experimental measurements show that the carbon poloidal velocity can be an order of magnitude above the neo-classical estimate within the ITB. This significantly affects the calculated radial electric field and therefore, the E x B flow shear used for example in transport simulations. Both the Weiland model and GLF23 reproduce the onset, location and strength of the ITB well when the experimental poloidal velocity is used while they do not predict the formation of the ITB using the neo-classical poloidal velocity in time-dependent transport simulation. The most plausible explanation for the generation of the anomalous poloidal velocity is the turbulence driven flow through the Reynolds stress. Both CUTIE and TRB turbulence codes show the existence of an anomalous poloidal velocity, being significantly larger than the neo-classical values. And similarly to experiments, the poloidal velocity profiles peak in the vicinity of the ITB and seem to be dominantly caused by flow due to the Reynolds stress. However, it is important to note that both the codes treat the equilibrium in a simplified way and this affects the geodesic curvature effects and geodesic acoustic modes (GAMs). Therefore, the results should be considered as indicative, and most probably provide an upper bound of the mean poloidal velocity as results from other codes including GAM dynamics show that they often serve as a damping mechanism to flows.