Methods of three-dimensional deconvolution with a point-spread function as frequently employed in optical microscopy to reconstruct true three-dimensional distribution of objects are extended to holographic reconstructions. Two such schemes have been developed and are discussed: an instant deconvolution using the Wiener filter as well as an iterative deconvolution routine. The instant 3d-deconvolution can be applied to restore the positions of volume-spread objects such as small particles. The iterative deconvolution can be applied to restore the distribution of complex and extended objects. Simulated and experimental examples are presented and demonstrate artifact and noise free three-dimensional reconstructions from a single two-dimensional holographic record.