Digital breast tomosynthesis (DBT), an emerging imaging modality, provides a pseudo-3D image of the breast. Algorithms to aid the human observer process these large datasets involve two key tasks: reconstruction and registration. Previous studies separated these steps, solving each task independently. This can be effective if reconstructing using a complete set of data, e.g., in cone beam CT, assuming that only simple deformations exist. However, for ill-posed limited-angle problems such as DBT, estimating the deformation is complicated by the significant artefacts associated with DBT reconstructions, leading to severe inaccuracies in the registration. In this paper, we present an innovative algorithm, which combines reconstruction of a pair of temporal DBT acquisitions with their simultaneous registration. Using various computational phantoms and in vivo DBT simulations, we show that, compared to the conventional sequential method, jointly estimating image intensities and transformation parameters gives superior results with respect to reconstruction fidelity and registration accuracy.