We investigated the effect of autonomous underwater vehicle (AUV) dynamics and navigation on underway submarine gravimetry. Our research was motivated by the need to obtain spatially dense marine gravity measurements close to the source of subkilometer-scale geologic features in the shallow oceanic crust. Such measurements have been previously obtained, for instance, with piloted submarines and towed sleds; however, the high cost and, in the case of on-bottom measurements, poor spatial sampling preclude routine acquisition of these measurements. Continuous underway gravity surveys with AUVs is a compelling cost-effective option, but this method requires separating the AUV accelerations from the measured gravity. We show that AUVs with a large distance between the center of buoyancy and the center of gravity have lower vertical accelerations than torpedo-shaped AUVs and consequentially are better suited for underway gravity surveys. Furthermore state estimators, which combine sensor measurements and models of the vehicle’s motion, provide superior estimates of the vehicle’s vertical accelerations than methods used in previous underway submarine gravity surveys. We simulated the use of these navigation methods in detecting dike swarms at the East Pacific Rise. Analysis showed that we can shorten filters used in reducing gravity data and consequentially provide improved measurements of the free-water anomaly with a minimal detectable spatial wavelength approximately 65% lower than previously reported results.