There were fifteen candidates in the AES selection and I suppose that Rijndael wasn't the single good algorithm. I know that Camellia is an evolution of E2 candidate and can be used as symetric cryptographic routines in TLS. Which others ex-candidates have today a "serious" implementation and a real usage ?
Well, not all of the AES-candidates have been considered good. In fact one of them even got broken at the presentation (within 20min!).
I'll list any (non-author) occurence of algoriths I'm aware of. Note: This list doesn't claim to be complete and may be extended by other people.
Now for a list of algorithms, for which I know there're an option
- SAFER, well it's predecessor, SAFER-K and SAFER-SK are available at least in the Crypto++ library
- CAST-256, available in the Crypto++ library
- MARS, finalist, available in the Crypto++ library, because of the finalist status it may be an option in other programs
- RC6, finalist, available in the Crypto++ library, available in Jetico's encryption tool, may be an option in other places as well as it's simpler than MARS and also has finalist status, according to wikipedia it was used pre-AES by the NSA. Furthermore usage of this algorithm can be costly as RSA Labratories seem to hold patents on it, as RC6 didn't win the AES-competition.
- Serpent, finalist, 2nd place of the AES process, back then it was believed to be the most secure available candidate, adopted in Crypto++ and available in Jetico's tools and in TrueCrypt and it's derivatives (e.g. CipherShed and VeraCrypt).
- Twofish, finalist, 3rd place of the AES process, back then it was approximiately as fast as Rijndael but it was believed to be more secure, adopted in Crypto++, TrueCrypt, by Jetico and in KeePass (1.x, 2.x as plug-in), and is part of the OpenPGP standard.