This paper addresses the problems of studying slow chemical exchange in spi
n systems that show scalar coupling. Line broadening due to exchange in thi
s regime is minimal, so rates are often obtained by selective-inversion exp
eriments. These are modified inversion-recovery experiments, in which only
one part of the exchanging system is inverted. These spins can then return
to equilibrium by exchange with the non-inverted sites, as well as by norma
l spin-lattice relaxation processes. This is a standard method for spin sys
tems which show little or no scalar coupling. The problem is more complex f
or coupled spin systems, because the longitudinal magnetizations are more d
ifficult to measure. There is no longer a one-to-one correspondence between
the longitudinal magnetizations and the observable lines in the spectrum.
The spectra depend on the flip angle of the observe pulse, there may be lon
gitudinal magnetizations that can not be measured reliably, and zero-quantu
m transitions may interfere. These problems are addressed here. The flip an
gle dependence can be exploited to enhance the information collected, and a
singular value decomposition (SVD) is used to obviate the numerical proble
ms of poorly determined longitudinal magnetizations. The role of the zero-q
uantum interference is also discussed. With these methods, coupled spin sys
tems can be treated like an uncoupled system. The technique is tested on tw
o molecules: ReBr(CO)(3)-2,6-pyridinedicarboxylate, a simple three-spin exc
hange system; and 4-nitroso-N,N-dimethylaniline, a four-spin system.