It is generally observed that the performance of double-resonance multiple-pulse-based (J) cross polarization (DCP) is superior to that of pulsed-free-precession (INEPT)-based sequences for net transfer of coherence between scalar coupled spins. Here, effects of relaxation and radiofrequency field inhomogeneity on transfer efficiency are analyzed for both methods. It is found that relaxation differences are relatively small between INEPT and DCP. RF inhomogeneity effects were found to significantly favor DCP over INEPT, contributing to the observed experimental differences in performance between the two methods. The differences suggest that triple-resonance cross polarization (TCP) between three coupled spins should yield better results than analogous INEPT-based net coherence transfers. The possibilities of TCP are theoretically analyzed by deriving the transfer functions for this type of experiment. It is found that the TCP transfer efficiency is low except in the case of equal scalar couplings. To widen the applications of the potentially interesting TCP method, a scheme involving a concatenation of triple- and double-resonance (\mathrm{CP}) is introduced (concatenated (\mathrm{CP}) or CCP). It is theoretically shown that such a sequence can be tuned to achieve complete in-phase transfer for all ratios of scalar couplings. The transfer times in this scheme are shown to be somewhat shorter than those required for optimally concatenated INEPT in-phase transfers. The transfer efficiencies of CCP were verified with a 3D HACA (N) NH experiment, in which the CA(\mathbf{N}-(\mathbf{N}) \mathbf{H}) transfer is driven by the CCP scheme. The experiment was carried out with ({ }^{15} \mathrm{~N},{ }^{13} \mathrm{C})-labeled T4 lysozyme ( (19 \mathrm{kDa}) ). The CCP experiment has much higher sensitivity than a version where the (\mathrm{CA}-\mathrm{N}-(\mathrm{N}) \mathrm{H}) transfer is driven by an INEPT scheme. ci 1995 Academic Press. Inc.