The main object of this paper is to emphasize that clouds—the nonprecipitating component of condensed atmospheric water—can produce a strong attenuation at operational microwave frequencies, although they present a low reflectivity preventing their radar detection. By way of a simple and realistic model, simulations of radar observations through warm precipitating targets are thus presented in order to quantify cloud attenuation. Simulations concern an airborne radar oriented downward and observing precipitation at four frequencies: 3, 10, 35, and 94 GHz. Two cases are first considered: a convective cell (vigorous cumulus congestus plus rain) and a stratiform one (nimbostratus plus drizzle) superimposed on the previous one. Other simulations are then performed on different types of cumulus (congestus, mediocris, and humilis) with various thicknesses characterized, in a microphysical sense, by their maximum liquid water content. Simulations confirm the low cumulus reflectivity ranging from −45 dBZ for the weakest cumulus (i.e., the humilis one) to −5 dBZ for the strongest one (i.e., the vigorous cumulus congestus). It reaches −35 dBZ for the nimbostratus cloud. On the other hand, cumulus attenuation [precisely path-integrated cloud attenuation (PICA)] is not negligible and, depending on the frequency, can be very strong: the higher the frequency, the stronger the PICA. At 3 GHz, the far less attenuated frequency, PICA for the vigorous cumulus congestus alone in the convective cell (embedded into the stratiform background) is on the order of 1.2 dB (1.5 dB) at 10 GHz, 16 dB (20 dB) at 35 GHz, and 80 dB (100 dB) at 94 GHz. For weaker cumulus, PICA is lower but, in certain cases, significant. All these results mean that it is necessary to be very careful about radar measurements if reliable information on precipitation—for example, the precipitation rate R—has to be deduced, particularly at high operational frequencies.