The publicly available FG-Net aging database is commonly used in many works for age estimation in order to evaluate performance. The FG-NET Aging Database contains 1,002 face images from 82 subjects.
 The Images of Groups Dataset is a collection of people images from Flickr images. 
They labeled each face into seven age categories: 0-2, 3-7, 8-12, 13-19, 20-36, 37-65, and 66+
http://chenlab.ece.cornell.edu/people...
 The MORPH database contains 1,724 face images of 515 individuals. These images represent a
diverse population with respect to age, gender, and ethnicity.
http://www.researchgate.net/profile/K...
 See also http://www.faceaginggroup.com/morph/ 
 See also ega-database as  StevenPuttemans suggested in the replies
 EDIT:
 There is also a recent article Age and Gender Classification using Convolutional Neural Networks by #gillevi  that uses a new aging database (Adience benchmark) (thanks!!!)
 http://www.openu.ac.il/home/hassner/p... http://www.openu.ac.il/home/hassner/p... http://www.openu.ac.il/home/hassner/A...
 The Adience set consists of images automatically uploaded to Flickr from smart-phone devices.
The entire Adience collection includes roughly 26K images of 2,284 subjects.