Quantitative estimation of individual biologic components in relaxation curves obtained in vivo may increase the specificity of tissue characterization by magnetic resonance imaging. In this study, the potential of biexponential curve analysis was evaluated in T1 and T2 measurements on the human brain at 1.5 tesla. Optimal experimental conditions were carefully observed, including the use of long TR values and a very small voxel size. T1 determination was based on a 12-points partial saturation inversion recovery pulse sequence. T2 determination involved a multiple spin echo sequence with 32 echoes. No genuine biexponentiality was demonstrated in the T1 and T2 relaxation processes of white matter, cortical grey matter, or cerebrospinal fluid. Thus, a monoexponential model seems adequate for description of the relaxation behaviour in these cases. Furthermore, the results suggest that the accuracy of the measurements may not be critically dependent on the voxel size employed.