The main namespaces/ontologies I use are as follows:

Dublin Core

Pretty much everyone uses this for annotating basic metadata such as author, creator, creation date. There are two main namespaces - the basic elements namespace:
dc: http://purl.org/dc/elements/1.1/
And the terms namespace:
dct: http://purl.org/dc/terms/

The elements namespace is simpler and doesn't specify domains and ranges for the properties, the terms namespace adds various additional properties and does specify this.

SKOS

The W3C's simple knowledge organisation system (Currently a Candidate Recommendation http://www.w3.org/TR/skos-reference/skos.html) is very useful for doing simple categorisation. And since it has properties that allow you to specify that categories are narrower/broader than others you can use some simple inference to select all the things categorised as within a given category or subcategory thereof. This is used widely on things like DBPedia and BBC linked data

skos: http://www.w3.org/2004/02/skos/core#

I find it is often very useful to do faceted browsing on properties which have values from a set of SKOS categories which are defined for your dataset.

ChangeSet

The Talis Changeset Ontology (http://purl.org/vocab/changeset) is an ontology for describing changes to an RDF document in terms of additions and removals of Triples. This is used by the Talis platform as part of their Changeset protocol (http://n2.talis.com/wiki/Changeset_Protocol) which allows you to specify modifications to your Talis stores using an RDF Graph encoded in this ontology. I've also used this in my own work as it provides a nice clear way of describing sets of changes to an RDF Graph, most importantly it allows you to link a Changeset to a previous Changeset to provide a version history of an RDF Graph.

cs: http://purl.org/vocab/changeset/schema#

Programmes Ontology

The Programmes Ontology from the BBC (http://purl.org/ontology/po/) is the ontology used by the BBC to describe their programming output. I use this a lot as this is the main dataset I run experiments on since it has characteristics which are useful to my research.

po: http://purl.org/ontology/po/

Vocab Status

The Vocab Status ontology (http://www.w3.org/2003/06/sw-vocab-status/ns) is a very simple ontology that can be used to mark up other ontologies to indicate the status of terms within them. This can be useful for a human reading an RDF serialization of your ontology to understand which properties and classes it is safe to make assumptions about and which may change in the future and should be treated with more caution.

vs: http://www.w3.org/2003/06/sw-vocab-status/ns#

FOAF

http://xmlns.com/foaf/0.1/

Together with Dublin Core, this is probably the best known and most widely used ontology. As well as relating people's identifiers and names, I often use foaf:depiction for images, foaf:primaryTopic and foaf:topic to relate a document to what it's about.

Geonames

http://www.geonames.org/ontology#

For places, geographical features etc. Links in to the comprehensive Geonames database.

DBpedia

http://wiki.dbpedia.org/Ontology

Since DBpedia is the de facto standard for finding URIs for many things, then sometimes it's handy to use its ontology too.

Event Ontology

http://motools.sourceforge.net/event/event.html

A flexible and relatively simple ontology for describing the what/when/where/who of events.

Just adding to the ones already mentioned (all of which I support as well!)...

SIOC

http://sioc-project.org/

This ontology is quite important if you want to get interoperability between all sorts of discussion platforms like blogs, forums, mailing lists, IRC, instant messaging, ... - basically you have a common model to describe all those in the same way (plus specialisation if you need it) and thus you can bridge between them and interlink discussions and communities. But it doesn't stop at discussions, you can describe all sorts of online content or items in general as well. It's very Web-oriented so it might be a bit flat for some advanced purposes (as compared to other models for describing media like FRBR). But it's also nicely extensible and can be used as the basis for many different things (for example I built an ontology for collaborative editing and reviewing on top of it).

There are many other ontologies I could mention but I'm not allowed to provide more than one link yet. :-)