Measurements of the transverse relaxation time T2 are usually conducted with the Carr Purcell Meiboom Gill (CPMG) pulse sequence, which causes T2-weighted magnetization. Diffusion effects are a common source of error in measurements of this kind, because the incoherent motion of spins in external magnetic field gradients distorts T2 weighting of the transverse magnetization. As a result, inaccurate T2-values are obtained. In this work, we present a method which completely compensates for the effect of diffusion.