Correlation reflectometry was first envisaged as a means to characterize density fluctuations in the JET tokamak. In radial correlation reflectometry, the radial scale of turbulence is estimated from the variation of coherence with the radial separation between the cutoff positions of two distinct probing waves. Currently, the JET correlation reflectometry diagnostic has four reflectometer systems, each one equipped with a fixed-frequency channel and a variable-frequency one [Hacquin 04], allowing measurements around four different radial positions. Recently, after the installation of low attenuation corrugated waveguides, good quality data with improved signal to noise ratio became available, as well as the software tools necessary for its routine analysis. From the physical point of view, correlation reflectometry measurements implicate not only the correlation length but also the level of turbulence. The correlation length of the microwaves, L, calculated from the reflected waves of the fixed and variable channels, must be corrected in terms of the level of density fluctuations to become a reliable estimate of the actual turbulence correlation length. It has been shown that the microwave correlation length is generally smaller than the turbulence's. Recent studies explain this behaviour with nonlinear effects caused by even relatively low turbulence levels. Without including such nonlinearity in theoretical models and simulations, measured correlation lengths might be expected to be larger than the true ones. This work reports on correlation reflectometry results obtained in recent JET experiments, in plasmas with formation of an Internal Transport Barrier (ITB).