There are historical discrepancies between empirical observations of Antarctic krill target strength and predictions using theoretical scattering models. These differences are addressed through improved understanding of key model parameters. The scattering process was modeled using the distorted-wave Born approximation, representing the shape of the animal as a bent and tapered cylinder. Recently published length-based regressions were used to constrain the sound speed and density contrasts between the animal and the surrounding seawater, rather than the earlier approach of using single values for all lengths. To constrain the parameter governing the orientation of the animal relative to the incident acoustic wave, direct measurements of the orientation of krill in situ were made with a video plankton recorder. In contrast to previous indirect and aquarium-based observations, krill were observed to orient themselves mostly horizontally. Averaging predicted scattering over the measured distribution of orientations resulted in predictions of target strength consistent with in situ measurements of target strength of large krill (mean length 40-43 mm) at four frequencies (43-420 kHz), but smaller than expected under the semi-empirical model traditionally used to estimate krill target strength.