They are roughly 1500 bp average for 16S and about 1800 bp for 18S. These aren't standard, so you can get some that differ a little in length, even within a single taxon. I assume you're asking about metagenomic uses, so for 454 you can get a solid chunk of them in a study with two primers and get a good assessment of diversity with specialized databases. More studies have investigated the 16S for 9 hypervariable regions of resolution (for example, see: Target Region Selection Is a Critical Determinant of Community Fingerprints Generated by 16S Pyrosequencing), than the 18S, but you should get fair resolution for both using 454 or Illumina at current sequencing lengths.

Please see: http://en.wikipedia.org/wiki/Ribosomal_RNA (paragraph "Prokaryotes vs. Eukaryotes").

I endorse what Josh has added... Though the lengths of 16s and 18s genes are around 1500 and 1800 respectively , getting them completely sequenced is tough considering the read lengths produced by various NGS platforms. However by targetting a very specific hypervariable region(say v3 or v6 in case of 16s) which are about ~180-220 bp in length, 454 or Illumina can definitely give a good throughput which can analysed various workflows that are available.