Relaxation rates in NMR are usually measured by intensity modulation as a function of a relaxation delay during which the relaxation mechanism of interest is effective. Other mechanisms are often suppressed during the relaxation delay by pulse sequences which eliminate their effects, or cancel their effects when two data sets with appropriate combinations of relaxation rate effects are added. Cross-correlated relaxation (CCR) involving dipole-dipole and CSA interactions differ from auto-correlated relaxation (ACR) in that the signs of contributions can be changed by inverting the state of one spin involved in the dipole-dipole interaction. This property has been exploited previously using CPMG sequences to refocus CCR while ACR evolves. Here we report a new pulse scheme that instead eliminates intensity modulation by ACR and thus allows direct measurement of CCR. The sequence uses a constant time relaxation period for which the contribution of ACR does not change. An inversion pulse is applied at various points in the sequence to effect a decay that depends on CCR only. A 2-D experiment is also described in which chemical shift evolution in the indirect dimension can share the same constant period. This improves sensitivity by avoiding the addition of a separate indirect dimension acquisition time. We illustrate the measurement of residue specific CCR rates on the non-myristoylated yeast ARF1 protein and compare the results to those obtained following the conventional method of measuring the decay rates of the slow and fast-relaxing (15)N doublets. The performances of the two methods are also quantitatively evaluated by simulation. The analysis shows that the shared constant-time CCR (SCT-CCR) method significantly improves sensitivity.