Young massive stars produce sufficient Lyman continuum photon luminosity PHI(i) to significantly affect the structure and evolution of the accretion disks surrounding them. A nearly static, ionized, isothermal 10(4) K atmosphere forms above the neutral disk for disk radii r < r(g) = 10(15)M1 cm, where M* = 10 M. M1 is the stellar mass. For r greater-than-or-similar-to r(g) the diffuse field created by hydrogen recombinations to the ground state in the photoionized gas above the disk produces a steady evaporation at the surface of the disk, and this H II gas flows freely out to the ISM (the '' disk wind ''). The detailed structure depends on the mass-loss rate M(w) of the fast, greater than or similar to 1000 km s-1, stellar wind from the massive star. A critical mass-loss rate M(cr) is defined such that the ram pressure of the stellar wind equals the thermal pressure of the H II atmosphere at r(g). In the weak stellar wind solution, M(w) < M(cr), the diffuse photons from the atmosphere above r(g) produce a photoevaporative mass-loss rate from the disk at r greater than or similar to r(g) of order 1 X 10(-5)PHI49(1/2)M1(1/2) M. yr-1, where PHI(i) = 10(49)PHI($() s-1. The resulting slow (10-50 km s-1) ionized outflow, which persists for greater than or similar to 10(-5) yr for disk masses M(d) approximately 0.3M*, may explain the observational characteristics of unresolved, ultracompact H II regions. In the strong stellar wind solution, ML(w) > M(cr), the ram pressure of the stellar wind blows down the atmosphere for r < r(g) and allows the stellar photons to penetrate to greater radii and smaller heights. A slow, ionized outflow produced mainly by diffuse photons is again created for r > r(g); however, it is now dominated by the flow at r(w) (>r(g)), the radius at which the stellar wind ram pressure equals the thermal pressure in the evaporating flow. The mass-loss rate from the disk is of order 6 x 10(-5) M(w-6)v(w8)PHI49(-1/2) M. yr-1, where M(w-6) = M(w)/10(-6) M., yr-1 and V(w8) = v(w)/1000 km s-1 is the stellar wind velocity. The resulting outflow, which also persists for greater than or similar to 10(5) yr may explain many of the more extended (r greater than or similar to 10(16) cm) ultracompact H II regions. Both the weak-wind and the strong-wind models depend entirely on stellar parameters (PHI(i), M*, M(w)) and are independent of disk parameters as long as an extended (r much greater than r(g)), neutral disk exists. We compare both weak-wind and strong-wind model results to the observed radio free-free spectra and luminosities of ultracompact H II regions and to the interesting source MWC 349.
