We have analyzed the available polarization surveys of the Galactic emission to estimate to what extent it may be a serious hindrance to forthcoming experiments aimed at detecting the polarized component of Cosmic Microwave Background (CMB) anisotropies. Regions were identified for which independent data consistently indicate that depolarization must be small. The power spectrum of the polarized emission, in terms of antenna temperature, was found to be described by $C_{ell}simeq (1.2pm 0.8)cdot 10^{-9}cdot (ell / 450)^{-1.8pm 0.3}cdot ( u/ 2.4{rm GHz})^{-5.8}$ K$^{2}$, from arcminute to degree scales. Data on larger angular scales ($ellle 100$) indicate a steeper slope $sim ell^{-3}$. We conclude that polarized Galactic emission is unlikely to be a serious limitation to CMB polarization measurements at the highest frequencies of the MAP and {sc Planck}/LFI instruments, at least for $ellge 50$ and standard cosmological models. The weak correlation between polarization and total power and the low polarization degree of radio emission close to the Galactic plane, found also in low-depolarization regions, is interpreted as due to large contributions to the observed intensity from unpolarized sources, primarily strong HII regions, concentrated on the Galactic plane. Thus estimates of the power spectrum of total intensity at low Galactic latitudes are not representative of the spatial distribution of Galactic emission far from the plane. Both total power and polarized emissions show highly significant deviations from a Gaussian distribution.