The 'shear-lag' analysis method is frequently used for analysis of str
ess transfer between the fiber and the matrix in composites. The accur
acy of shear-lag methods has not been critically assessed, in part bec
ause the assumptions have not been fully understood. This paper starts
from the exact equations of elasticity for axisymmetric stress states
in transversely isotropic materials and introduces the minimum assump
tions required to derive the most commonly used shear-lag equations. T
hese assumptions can now be checked to study the accuracy of shear-lag
analysis on any problem. Some sample calculations were done for stres
s transfer from a matrix into a broken fiber. The shear-lag method did
a reasonable job (within 20%) of predicting average axial stress in t
he fiber and total strain energy in the specimen provided the shear-la
g parameter most commonly used in the literature is replaced by a new
one derived from the approximate elasticity analysis. The shear-lag me
thod does a much worse job of predicting shear stresses and energy rel
ease rates. Furthermore, the shear-lag method does not work for low fi
ber volume fractions. (C) 1997 Elsevier Science Ltd.