For tiny structures (I did neonatal rat tongue, and chicken embryo hindbrain) I had good success using a peltier in the cryostat. In addition, I would use for larger structures dry ice cooled iso-pentane to snap freeze the specimen. TissueTek cryomolds were fabulously easy to use. This allowed me to orient my specimen easily, and give me an easy-to-grip part of the mold to hold to dip (but not immerse) into the isopentane. Don't forget to cryoprotect with sucrose like Mohamed said.
you don't *have* to cryoprotect, but if you don't you'll need to flash freeze. The success of how fast your flash freeze is. thick/tough specimens may not flash freeze rapidly enough. In addition, for fluorescent imaging in particular, you don't want to overfix your specimens either. You may get even better results with less post-fix time.