Velocity profiles in Couette flow are measured in a wormlike micellar solution made of cetyltrimethylammonium bromide (CTAB), sodium salicylate (NaSal), and water, at R (= [NaSal]/[CTAB]) = 2 and at R = 4; [CTAB] = 100 mM. Velocity profiles were obtained by using a two-incident beam laser Doppler technique. Profiles reveal that one of the micellar solutions (R = 2) becomes heterogeneous a long time after flow inception, even at very low imposed shear rates. However, profiles do not correspond to what is expected for gradient shear banding, because the fluid splits in one section close to the moving cylinder where the local mean velocity depends linearly on the gap position and in a second section presenting important velocity fluctuations. Close to the static cylinder, there is a third section where the fluid does not flow; it behaves like a slipping block. On the other hand, at high imposed shear rates, the former slipping block flows and presents a linear profile. Here, velocity profiles are consistent with gradient shear banding. The onset of shear banding was observed. The picture of two stable shear bands separated by a thin steady interface is not always valid. Inhomogeneous flow could be observed, although it cannot be classified as shear banding. In addition, conditions can be found where, as shear rate is increased and before shear banding appears, instead of a thin interface, a fluctuating intermediate band can be observed. On the contrary, for the R = 4 solution, the flow never becomes heterogeneous even at high shear rates. Flow curves were measured in a Couette cell under shear rate control in two cases, when stress is sensed with the moving cylinder and when it is sensed with the static cylinder of the cell. Differences between the flow curves can be explained by using the velocity profiles for both solutions.