The spectral characteristics of anvils in tropical areas (25°S–25°N) have been investigated on the basis of data from the tropical rainfall measuring mission’s (TRMM) precipitation radar (PR) and a visible and infrared scanner (VIRS), from 1998 to 2007. The anvils’ vertical structures were captured by TRMM PR and categorized into two subtypes: ice anvils with an echo base of ≥6 km and mixed anvils with an echo base between 3 and 6 km. Visible and infrared signals for the anvils, which are from reflectance at 0.63 and 1.6 μm (hereafter referred to as RF1 and RF2, respectively) and the equivalent brightness temperatures of a black body at 3.7, 10.8, and 12.0 μm (hereafter referred to as TB3, TB4, and TB5, respectively), were derived simultaneously by use of TRMM VIRS. The findings reveal that the normalized frequency distribution (i.e., probability density functions, PDF) of anvil tops and bases have a bimodal distribution whereas that of anvil thickness has a single-peak curve. For visible signals, the PDF distribution of RF1 (RF2) for anvils, mixed anvils, and ice anvils has an approximately symmetric distribution with a tropics-wide averages of 0.74, 072, and 0.80 (0.21, 0.21, and 0.20), respectively. It can be concluded that ice anvils are optically thicker and contain many more ice-cloud droplets at the cloud top than mixed anvils. RF1 of anvils is usually lower over land than over ocean, by ~0.1, whereas RF2 of anvils is usually higher over land than over ocean, by ~0.3. This implies that anvil clouds have thinner optical depth and their cloud tops consist of many more small ice droplets over land than over ocean. For infrared signals, TB4 is regarded as a representative channel. The PDF distribution of TB4 for anvils and mixed anvils is broad, with tropics-wide averages of 229.2 and 232 K, respectively. They contain two peaks and the secondary peak lies at a much lower value. For ice anvils, the PDF distribution of TB4 is a single-peak curve with a tropics-wide average of 219.5 K. Mean TB4 of anvils, mixed anvils, and ice anvils is usually lower over land than over ocean, by ~3.6 K, ~1.6 K, and 5 K, respectively. In addition, land–ocean differences between RF1-to-RF2 reflectance ratios are more obvious than those between TB4 and TB5 brightness temperatures. Furthermore, these spectral signals have been proved to be a potentially useful way of revealing the presence of anvils over land and ocean, and especially for separating anvils into the mixed and ice subtypes. Finally, long-term mean values reveal that the spectral signals of anvils and of stratiform and convective precipitating clouds are indeed different, and that the difference varies with different geographical location. In real time, however, uncertainty still exists when only spectral signals are used to discriminate anvils from stratiform and convective precipitating clouds, which means active satellite observations are indispensable.