Procedures are described for significantly improving the sensitivity of the recently proposed TROSY (transverse relaxation-optimized spectroscopy) experiment (K. Pervushin et al., 1997, Proc. Natl. Acad. Sci. USA 94, 12366-12371). The TROSY experiment takes advantage of destructive interference between dipolar and chemical shift anisotropy relaxation mechanisms to achieve substantial reductions in resonance linewidths in heteronuclear correlation spectra; the effect is significant particularly for studies of large molecular weight systems at very high static magnetic field strengths. A (square root 2) improvement in the sensitivity of the TROSY experiment is achieved by implementation of the PEP (preservation of equivalent pathways) scheme (J. Cavanagh and M. Rance, 1990, J. Magn. Reson. 88, 72-85). An additional significant improvement in sensitivity for 15N-labeled samples in H2O solution is realized through a simple modification of the 1H-15N TROSY pulse sequence to return the water magnetization to its equilibrium position (+z axis) at the beginning of the acquisition period. Relaxation-induced imbalance between the coherence transfer pathways utilized in the TROSY refocusing period is shown theoretically and experimentally to give rise to additional unanticipated signals in TROSY spectra.