This paper argues that star-forming environments should be classified into
finer divisions than the traditional isolated and clustered modes. Using th
e observed set of Galactic open clusters and theoretical considerations reg
arding cluster formation, we estimate the fraction of star formation that t
akes place within clusters. We find that less than similar to 10% of the st
ellar population originates from star-forming regions destined to become op
en clusters, confirming earlier estimates. The smallest clusters included i
n the observational surveys (having at least N similar to 100 members) roug
hly coincide with the smallest stellar systems that are expected to evolve
as clusters in a dynamical sense. We show that stellar systems with too few
members N < N-* have dynamical relaxation times that are shorter than thei
r formation can be (<similar to>1-2 Myr), where the critical number of star
s N-* approximate to 100. Our results suggest that star formation can be ch
aracterized by (at least) three principal modes: (I) isolated singles and b
inaries, (II) groups (N < N-*), and (III) clusters (N >N-*). Many-if not mo
st-stars form through the intermediate mode in stellar groups with 10 < N <
100. Such groups evolve and disperse much more rapidly than do open cluste
rs; groups also have a low probability of containing massive stars and are
unaffected by supernovae and intense ultraviolet radiation fields. Because
of their short lifetimes and small stellar membership, groups have relative
ly little effect on the star formation process (on average) compared to lar
ger open clusters.