Water proton longitudinal relaxation has been measured in agar and cross-linked bovine serum albumin (BSA) using modified selective excitation (Goldman-Shen and Edzes-Samulski) pulse sequences. The resulting recovery curves are fit to biexponentials. The fast recovery rate gives magnetization transfer (MT) information, which is complementary to that given by steady-state saturation methods. This rate provides an estimate of the strength of the coupling of the immobile proton pool to the mobile proton pool. Near their optimal pulse power values, the Goldman-Shen and Edzes-Samulski sequences give fast recovery rates that agree with each other. However, these measured fast recovery rates are dependent on the pulse power, an effect not predicted by the coupled two-pool model. For 8% agar and 17% BSA, both methods (at optimal pulse powers) give rates in the neighborhoods of 210 and 64 Hz, respectively. The Goldman-Shen and Edzes-Samulski pulse sequences have several advantages over those techniques based on steady state saturation: no long saturating pulses, shorter measurement time, and reduced necessity for making lineshape or fitting technique assumptions. The principle disadvantages are smaller effects on the NMR signal, less complete characterization of the MT system, and, in the case of the Goldman-Shen sequence, greater pulse power.