Long-slit spectroscopic observations of η Carinae and its Homunculus have recently been obtained using STIS on the Hubble Space Telescope (HST). We have extracted the spectrum of the central source using a 01×013 aperture. As expected, the spectrum is very different from ground-based spectra obtained under normal seeing conditions (i.e., 1'')—the HST observed spectrum exhibits primarily broad permitted emission lines (v ≈ 500 km s^-1), while ground-based spectra show strong narrow (v ≈ 40 km s^-1) permitted and forbidden lines (primarily [Fe II], Fe II, and H I) superimposed on a broad emission-line spectrum. In the optical, the spectrum shows strong H I, He I, and Fe II emission lines, many exhibiting P Cygni absorption components. Emission features due to N I, Si II, Na I, Mg II, Ca II, and Al II can also be identified. Only a few weak broad lines of [Fe II] are seen. The spectrum is qualitatively similar to a much less luminous object, the extreme P Cygni star HDE 316285. We have performed a detailed analysis of the spectrum using the non-LTE line blanketed wind code of Hillier & Miller. Despite the complexity of η Carinae, we are able to obtain a good fit to the optical emission line spectrum using a model with a mass-loss rate of 10^-3 M yr^-1 . The weakness of the electron-scattering wings indicates that the wind is clumped, with a volume filling factor of approximately 0.1. Because of the parameter range and the extremely dense wind associated with the central star, our best-fit model is not unique. As is the case for HDE 316285, there is a strong coupling between the derived mass-loss rate and the derived N(H)/N(He) abundance ratio. In addition, the wind is so dense that the star's surface cannot be observed. Consequently, the effective temperature of the underlying star is not well determined. Because of the rich emission-line spectrum we are able to place limits on many abundances: Mass fractions of species such as Fe and Mg (and perhaps Ca, Si, and Al) are solar to within a factor of 2. In accord with standard evolutionary scenarios, Na is slightly enhanced, the N mass fraction is at least a factor of 10 over solar, while C and O show substantial depletions. The adopted luminosity of 5 × 10⁶ L is based on the observed IR flux and an assumed distance of 2.3 kpc. Thus, based on the Eddington limit, the minimum mass of the system is approximately 120 (L/5 × 10⁶ L) M [for N(H)/N(He) = 5]. While η Carinae may be a binary, recent evidence suggests that the purported secondary star has a mass less than 30 M. Thus the primary star is currently more massive than 90 M, and the initial mass of the star should conservatively have been in excess of 150 (L/5 × 10⁶) M.