In this poster, we examine the science potential of an 800 meter interferometer such as the OHANA Array. The working assumptions are a K = 12 limiting magnitude, a 0.5 milliarcsecond resolution at K band, and a small (diffraction limit of individual telescope) field of view. The science cases described herein are by no means exhaustive and perhaps not even the ones that will eventually be carried out, but serve to illustrate the potential of the array. We expect that operation of the array will be proposal driven, so the actual science will come from the Mauna Kea communities. Our philosophy is that any measurement that can be made at a dedicated interferometer facility should not be a strong driver for OHANA. Therefore the science areas discussed in the poster focus on very high angular resolution measurements of faint sources. In some cases, science which can be addressed with simpler or dedicated facilities at an exploratory level can be carried to a significant new capability with OHANA. A limitng magnitude of 12 was obtained by simple computations, but first tests on the sky with the injection module (See adjacent poster on Phase I) will help narrow down this figure. At such sensitivity, Cepheid pulsations can be studied in considerable detail for a wide range of stellar parameters, leading to enhanced confidence in the accuracy of their use for distance measurement with minimal extrapolation or inferrence. The disk/star interaction zone in young stellar objects can be resolved with unprecedented detail for a range of masses and ages, providing direct information about the jet formation region, accretion rates and disk conditions. The broad line region of active galactic nuclei can be studied in a large number of sources of differing characteristics, testing specific models for AGN nuclear structure. For OHANA Phase III, a dual-star phase tracking capability is planned. With the resulting increased sensitivity, direct brown dwarf diameter measurement will provide a strong check on evolution models. Microlensing events could be resolved and provide unique new information about the lensing and the lensed objects.