The functions of mitochondria and chloroplasts rely on thousands of proteins, mostly imported from the cytosol through specialized import channels. Neither the detailed import mechanisms nor the identities of all targeted proteins are known. Recent surprises include unexpected results concerning import receptors, unexpectedly frequent dual-targeting of proteins, and the discovery of novel routes of protein trafficking. Such findings make it more difficult to predict which proteins really are targeted to organelles. By combining experimental and bioinformatics data, we estimate the size of the mitochondrial and plastid proteomes to be approximately 2000 and 2700 proteins, respectively. Advances in cell and organelle fractionation coupled with modern proteomics techniques are probably the best route to understanding organellar protein composition.