Uniprot contains 542000 protein sequences, and two orders of magnitude more including TrEMBL. The PDB has just under 100000 structures, so taking these at face value your list would contain between 400 thousand and 40 million entries! These numbers are rough as there is redundancy in both sequence and structure databases.

The challenge in finding a target for structure determination is to find one which is both tractable and interesting.

You also might examine the TargetTrack database which summarizes the results from the U.S. Protein Structure Initiative (http://sbkb.org/tt/). For example, you could find proteins that have been successfully purified or even crystallized, but no structure was determined. (You would want to double check with the PDB to make sure that some other group didn't determine the structure independently.) You can search TargetTrack by sequence, Pfam domain, organism, experimental status, etc.

As someone else posted, there are millions of sequences for which no exact 3D structure is published, and hundreds of thousands with no detectable homology to a known structure. So depending on what you want to do, it would probably be good to narrow your search.