Background and Objectives. Important issues regarding the spread of solutio
ns in the epidural space and the anatomy of the site of action of spinal an
d epidural injections are unresolved. However, the detailed anatomy of the
spinal canal has been incompletely determined. We therefore examined the mi
croscopic anatomy of the spinal canal soft tissues, including relationships
to the canal walls. Methods. Whole mounts were prepared of decalcified ver
tebral columns with undisturbed contents from three adult humans. Similar m
aterial was prepared from a macaque and baboon immediately on death to cont
rol for artifact of tissue change after death. Other tissues examined inclu
ded nerve root and proximal spinal nerve complex and dorsal epidural fat ob
tained during surgery. Slides were examined by light microscopy at magnific
ations of 10-40x. Results. There is no fibrous tissue in the epidural space
. The epidural fat is composed of uniform cells enclosed in a fine membrane
. The dorsal fat is only attached to the canal wall in the dorsal midline a
nd is often tenuously attached to the dura. The dura is joined to the canal
wall only ventrally at the discs. Veins are evident predominantly in the v
entral epidural space. Nerve roots are composed of multiple fascicles which
disperse as they approach the dorsal root ganglion. An envelope of arachno
id encloses the roots near the site of exit from the dura. Conclusions. The
se features of the fat explain its semifluid consistency. Lack of substanti
al attachments to the dura facilitate movement of the dura relative to the
canal wall and allow distribution of injected solution. Fibrous barriers ar
e an unlikely explanation for asymmetric epidural anesthesia, but the midli
ne fat could impede solution spread. Details of nerve-root structure and th
eir envelope of pia-arachnoid membrane may be relevant to anesthetic action
.