We report on the detection of optically thick free-free radio sources in th
e galaxies M33, NGC 253, and NGC 6946 using data in the literature. We inte
rpret these sources as being young embedded star birth regions that are lik
ely to be clusters of ultracompact H II regions. All 35 of the sources pres
ented in this article have positive radio spectral indices (alpha > 0 for S
-v proportional to v(alpha)), suggesting an optically thick thermal bremsst
rahlung origin from the H II region surrounding the hot stars. The estimate
d emission measures for these sources are EM6cm greater than or similar to
10(8) cm(-6) pc, and energy requirements indicate that the sources in our s
ample have a range of a few to similar to 560 O7 V star equivalents powerin
g their H II regions. Assuming a Salpeter initial mass function with lower
and upper mass cutoffs of 1 and 100 M., respectively, this range in N-Lyc c
orresponds to integrated stellar masses of 0.1-60 x 10(3) M. For roughly ha
lf of the sources in our sample there is no obvious optical counterpart, wh
ich gives further support for their deeply embedded nature; for most of the
remaining sources, the correspondence to an optical source is insecure owi
ng to relative astrometric uncertainty. Their luminosities and radio spectr
al energy distributions are consistent with H II regions modeled as spheres
of plasma with electron densities from n(e) similar to1.5 x 10(3) to simil
ar to1.5 x 10(4) cm(-3) and radii of similar to1-7 pc. Because of the high
densities required to Dt the data, we suggest that the less luminous of the
se sources are extragalactic ultracompact H II region complexes, those of i
ntermediate luminosity are similar to W49 in the Galaxy, and the brightest
will be counterparts to 30 Doradus when they emerge from their birth materi
al. These objects constitute the lower mass range of extragalactic "ultrade
nse H II regions," which we argue are the youngest stages of massive star c
luster formation yet observed. The sample presented in this paper is beginn
ing to fill in the continuum of objects between small associations of ultra
compact H II regions and the analogous massive extra-galactic clusters that
may evolve into globular clusters.