The possibility of uniform horizontal correlation scales for tropical oceanic rainfall has been examined by a study of satellite-observed microwave data as a proxy measure of rain rates. From the brightness temperatures from the electrically scanning microwave radiometer on Nimbus 5 near nadir during the year of 1974, the mean spatial autocorrelation function as a function of simultaneous pixel separation (25 km pixel contiguous center-to-center spacing) was calculated in each 5¿ by 5¿ grid box over the tropical Pacific and Atlantic for each season. The equal-time spatially lagged correlations were compared for geographical dependence to investigate the hypothesis of homogeneity. A simple model of the spatial statistics of the microwave brightness temperatures was used, consisting of a mixture of uncorrelated spatial white noise incoherently superimposed on a spatially correlated field (spatial red noise). The red noise, signals are presumed to be generated by convective activity in the tropical atmosphere. The parameters of the red noise are consistent with this scheme over the tropical oceans, yielding a uniform spatial scale of about 50 km throughout. We believe the red noise field in the microwave data is due mainly to rainfall, but some contributions from cloud liquid water and water vapor amount in the atmosphere are also possible. Due to the inherent limitation of our approach, it is impossible to test the homogeneity hypothesis rigorously. However, there are no reasons to reject it, and the likelihood of a uniform spatial scale of tropical rainfall over oceans is enhanced through this study. ¿1991 American Geophysical Union