Since the launch of Hubble Space Telescope (HST) 9 yr ago, Cepheid distances to 25 galaxies have been determined for the purpose of calibrating secondary distance indicators. Eighteen of these have been measured by the HST Key Project team, six by the Supernova Calibration Project, and one independently by Tanvir. Collectively, this work sets out an array of survey markers over the region within 25 Mpc of the Milky Way. A variety of secondary distance indicators can now be calibrated, and the accompanying four papers employ the full set of 25 galaxies to consider the Tully-Fisher relation, the fundamental plane of elliptical galaxies, Type Ia supernovae, and surface brightness fluctuations. When calibrated with Cepheid distances, each of these methods yields a measurement of the Hubble constant and a corresponding measurement uncertainty. We combine these measurements in this paper, together with a model of the velocity field, to yield the best available estimate of the value of H0 within the range of these secondary distance indicators and its uncertainty. The uncertainty in the result is modeled in an extensive simulation we call the "virtual Key Project." The velocity-field model includes the influence of the Virgo cluster, the Great Attractor, and the Shapley supercluster, but does not play a significant part in determining the result. The result is H0 = 71 ± 6 km s-1 Mpc-1. The largest contributor to the uncertainty of this 67% confidence level result is the distance of the Large Magellanic Cloud, which has been assumed to be 50 ± 3 kpc. This takes up the first 6.5% of our 9% error budget. Other contributors are the photometric calibration of the WFPC2 instrument, which takes up 4.5%, deviations from uniform Hubble flow in the volume sampled (≲2%), the composition sensitivity of the Cepheid period-luminosity relation (4%), and departures from a universal reddening law (~1%). These are the major components that , when combined in quadrature, make up the 9% total uncertainty. If the LMC distance modulus were systematically smaller by 1 σ than that adopted here, the derived value of the Hubble constant would increase by 4 km s-1 Mpc-1. Most of the significant systematic errors are capable of amelioration in future work. These include the uncertainty in the photometric calibration of WFPC2, the LMC distance, and the reddening correction. A NICMOS study is in its preliminary reduction phase, addressing the last of these concerns. Various empirical analyses have suggested that Cepheid distance moduli are affected by metallicity differences. If we adopted the composition sensitivity obtained in the Key Project's study of M101, and employed the oxygen abundances measured spectroscopically in each of the Cepheid fields we have studied, the value of the Hubble constant would be reduced by 4% ± 2% to 68 ± 6 km s-1 Mpc-1.
