We compare the solar wind proton flux latitude dependence derived in the past from the interstellar neutral H distribution in the inner heliosphere from Lyman &agr; observations with recent in situ solar wind observations by Ulysses [Goldstein et al., 1995, Phillips et al., 1995, 1996, J. L. Phillips, private communication, 1996>. We find common features, such as a significant proton flux decrease with increasing heliographic latitude (about 30%) in the low-latitude regions and broad plateaus of low particle fluxes (2.0--2.5¿108 protons cm-2 s-1) around the poles. We use our model of interstellar H distribution under the influence of a multiparameter, latitude dependent solar wind to investigate the effects of a solar wind distribution matching as closely as possible the south heliographic Ulysses observations. For the first time, multiple scattering is included in such an anisotropic model. The Ulysses-type wind is found to produce a secondary minimum of Lyman &agr; intensity in the upwind direction, something already observed near solar minimum of activity. However, the modeled feature has a larger amplitude than the observed one, probably an indication of smoothing due to the combination of solar rotation and waviness of the neutral sheet. The total solar wind particle flux and the full Sun-averaged ionization rate of the interstellar neutral H are estimated in various cases. For identical equatorial and polar fluxes, the existence of broad plateaus results in a significant reduction of the average neutral H ionization (and then of the ionization cavity) when comparing with models using a classical harmonic dependence with latitude. As a result, the downwind cavity is less depleted. This may partially explain some discrepancies between the expected and observed Lyman &agr; emissions from the interplanetary hydrogen cavity, in particular, the excess of emission from the downwind cavity compared with the classical model.¿ 1997 American Geophysical Union