One- and two-jet inclusive quantities in hadron collisions have alread
y been calculated to next-to-leading order accuracy, using both the su
btraction and the slicing method. Since the one-loop corrections have
recently been obtained for all five-parton amplitudes, three-jet inclu
sive quantities can also be predicted to next-to-leading order. The su
btraction method presented in the literature is based on a systematic
use of boost-invariant kinematical variables, and therefore its applic
ation to three-jet production is quite cumbersome. In this paper we re
-analyze the subtraction method and point out the advantage of using a
ngle and energy variables. This leads to simpler results and has compl
ete generality, extending its validity to n-jet production. The formal
ism is also applicable to n-jet production in e(+)e(-) annihilation an
d in photon-hadron collisions. All the analytical results necessary to
construct an efficient numerical program for next-to-leading order th
ree-jet inclusive quantities in hadroproduction are given explicitly.
As a new analytical result, we also report the collinear limits of all
the two-to-four processes.