During the last five years or so, several new satellite precipitation products with hourly or 3-hourly resolution and on ~0.25deg grids have been produced by several groups at NASA GSFC, NOAA NCEP/CPC, and UC, Irvine. These products are often created by using the relatively reliable rain rates derived from microwave measurements to calibrate the estimates based on infrared observations from geostationary satellites, although the detailed techniques differ among the products. Because of their high temporal and spatial resolution and being near real-time, these merged satellite products could potentially have many applications. However, there have been few efforts to systematically evaluate and compare these products. Here we employ the coherent diurnal cycle as a test bed to evaluate these satellite products and compared with available surface observations over the U.S. and the globe. Preliminary results show that the satellite products show stronger diurnal amplitudes (in percentage of the daily mean) over most land areas than those in surface observations. Their diurnal phase (time of maximum) tend to lag that in surface observations by 2-4 hours over many land areas. Without inclusion of any surface observations in these pure satellite products, the geographic distributions of seasonal-mean precipitation amount also differ considerably among these products and from surface observations.