The characterization of the shape of liquid elements in a spray is a good way to analyze atomization processes. Experimental approaches based on common imaging techniques suffer from partial information given by 2D images to characterize intrinsically 3D objects such as liquid droplets. In this paper, we address the question, To what extent can the shape parameters measured on 2D images reveal the 3D content of the droplet shape? An analytical approach is first adopted to determine relationships between 2D and 3D shape parameters for two families of objects, the oblate and prolate spheroids. Two numerical databases obtained from direct numerical simulation of two-phase flows are then explored for their ability to give complete 3D shape information and extrapolate 2D parameter values from projections on planes selected according to flow characteristics. Focus is put on one shape parameter particularly sensitive to 3D to 2D projection effects, namely the uniformity parameter introduced by Blaisot and Yon (2005). Statistics obtained from the numerical databases are used to guide the analysis of results extracted from experimental images. It is shown that statistics on a 3D shape parameter could be induced from the ones for the 2D parameters. Two conditions are necessary: (1) the 2D projection is performed perpendicularly to the main flow direction, which is always the case in experiments; and (2) a particular care must be put on the determination of the statistics of orientations of the main axis of the liquid elements. This last point should be tackled in future experimental analyzes to estimate 3D shape parameter statistics from 2D images.