The analysis presented provides statistical information on the occurrence of gravity waves using the dense sampling geometry provided by the COSMIC/FORMOSAT3 constellation. This is achieved by deriving profiles of the gravity wave temperature perturbations from satellite observation and then identifying profile pairs; the magnitude of the cross S-transform derived from normalized profile pairs is then compared with Monte Carlo simulations to determine whether the same wave can be observed in both profiles. This procedure allows the percentage of paired observations which contain the same wave packet to be derived at a given significance level. This study shows that the occurrence of coherent waves varies strongly as a function of latitude and season. The latitudinal variation observed is demonstrated to be at least partially related to the preferential occurrence of long horizontal wavelength gravity waves at the equator. However, other factors obviously also have an impact, for example seasonal variations in the occurrence in the subtropics seem to be linked to critical level filtering above the subtropical winter jet. The use of the discussed methodology in identifying profile pairs which should be used in the calculation of gravity wave characteristics, such as horizontal wavelength, is also considered. Application of this technique as a quality control step seems to reduce biases introduced into the horizontal wavelength distribution when uncorrelated profile pairs are used to derive phase information. Thus, it is highly recommended to apply some statistical technique to identify profile pairs which include the same wave before applying schemes to derive horizontal wavelength and/or momentum fluxes.