From a separate thread, addressed here to avoid going off-topic.
I was told by Hal Brown (from the Delaware State MedEx Office) that Family Reference Samples (FRS) are processed entirely by an automated system, and any nucDNA samples taken from maternal relatives are also processed for mtDNA.
However, if only paternal relatives are available (e.g., Mindi Chambers), no mtDNA profile can be developed.
Also, if the unidentified body is too badly degraded, they might not be able to extract nucDNA even if they are able to extract mtDNA. As I understand it, mtDNA is more resistant to degradation than nucDNA.
Also, although it is often denied that there are known instances of false negatives in CODIS, I know of one instance where DNA was available in CODIS for both the MP and the UID, and a hit was not triggered.
Dawn Renee Higdon of Ridgecrest CA went missing in Aug'88 and her skeletal remains were found in Jul'90 in San Bernadino County.
Here is the link to info regarding Dawn (no info is available about the previously unidentified remains)
http://forthelost.wordpress.com/2010/08/02/dawn-higdon-missing-twenty-two-years/
These cases remained unmatched despite the existence of a single FRS from one of Dawn's relatives, and a DNA profile for the UID in CODIS. (I'm not sure whether they were nucDNA or mtDNA, or whether the single relative was maternal or paternal). It was only after additional relatives submitted FRS samples that a feature in CODIS called "Pedigree Tree" evaluated the multiple samples collectively and then was able to flag the match to the UID case.