(15)N spin relaxation data have provided a wealth of information on protein dynamics in solution. Standard R(1), R(1)(rho), and NOE experiments aimed at (15)N[(1)H] amide moieties are complemented in this work by HA(CACO)N-type experiments allowing the measurement of nitrogen R(1) and R(1)(rho) rates at deuterated (15)N[(2)D] sites. Difference rates obtained using this approach, R(1)((15)N[(1)H]) - R(1)((15)N[(2)D]) and R(2)((15)N[(1)H]) - R(2)((15)N[(2)D]), depend exclusively on dipolar interactions and are insensitive to (15)N CSA and R(ex) relaxation mechanisms. The methodology has been tested on a sample of peptostreptococcal protein L (63 residues) prepared in 50% H(2)O-50% D(2)O solvent. The results from the new and conventional experiments are found to be consistent, with respect to both local backbone dynamics and overall protein tumbling. Combining several data sets permits evaluation of the spectral density J(omega(D) + omega(N)) for each amide site. This spectral density samples a uniquely low frequency (26 MHz at a 500 MHz field) and, therefore, is expected to be highly useful for characterizing nanosecond time scale local motions. The spectral density mapping demonstrates that, in the case of protein L, J(omega(D) + omega(N)) values are compatible with the Lipari-Szabo interpretation of backbone dynamics based on the conventional (15)N relaxation data.