Monopole and dipole logging data at the Mallik 5L-38, Mackenzie Delta, Canada, provide a challenge for sonic velocity and attenuation models used to remotely estimate pore-space gas hydrate content. Velocity and attenuation are linked, with velocity dispersion causing increased attenuation. Sonic waveforms for Mallik 5L-38, however, show no velocity dispersion in gas hydrate-bearing layers, yet are highly attenuated. Attenuation models applied to Mallik 5L-38 data are shown to be inconsistent with the observed velocity measurements, and therefore are suspect in their ability to predict gas hydrate content. A model explicitly linking velocity and attenuation data is presented, accurately predicting gas hydrate content from velocity data alone while demonstrating that the attenuation mechanisms at the Mallik 5L-38 site have not yet been identified.