It's a good question, and if I remember from my undergrad courses and my casual encounters with stereochemistry since, you're right--not all compounds containing stereocenters are necessarily optically active.
This is a subtle but crucial point that needs to be drawn out better throughout the pages in which this occurs. "optically active" is used because it is an easily described macroscopically observable phenomenon, but a more abstract description at the atomic level requires discussion in terms of stereochemistry rather than optical activity.
consider, for example, the (still controversial) case of thalidomide, in which it is believed that only one enantiomer is teratogenic. the racemic mixture of thalidomide is not optically active, yet stereochemical concerns still exist for this mixture. Similarly for just about any other racemate--the unresolved, bulk racemate is not optically active in the sense that it doesn't rotate plane polarized light, but the stereocenters still exist.
With that said, in my opinion, we should instead have a page on stereochemistry which includes a discussion on optical activity, instead of the other way round, which we have now. But, I bring this up for discussion instead of just doing it so that we don't have a tug-of-war similar to what is happening with ascorbic acid and vitamin c--I'm not keen to meet the aggravation that such would entail.
Regarding thalidomide, I doubt very much that thalidomide was tested as an enatiomerically pure isomer but then manufactured and administered as the racemate--enantioselective synthesis is almost always more involved and therefore more expensive, and in many cases not possible or when possible, practical. without yet digging up sources, I expect that what happened was that it was developed as a racemate as well, and just not tested for teratogenic effects.
Note, for instance, in this account that the drug was synthesized in 1953, and marketed 1957-1962, but not enantioselectively tested until 1961:
http://www.chem.yale.edu/~chem125/125/thalidomide/thalidomide.html
consider also
http://yarchive.net/env/thalidomide.html
Perhaps thalidomide isn't a great example for optical activity or stereochemistry :-)
maybe the story about Pasteur laboriously separating tartrate salt crystals would be better?
I started the Optical isomerism page and agree that some of the info should be on a stereochemistry page - I don't think there'll be a tug of war. But since I don't actually no anything about stereochemistry compared to Joe above I'm probably not the man to do it.
I deleted the second part of the thalidomide passage as my understanding I think is wrong. Why would each enantiomer need to be tested separately if they had tested both together and found no problems?
--sodium
--sodium