There have been an increasing number of social tagging systems on the web that allow users to contribute tags collaboratively to describe various resources. As the number of tags continue to grow, it is important that we can describe and measure their quality. There have been conflicting opinions about the quality of user-contributed tags but the issue has not been systematically studied. In this work-in-progress paper, we propose a preliminary methodology for assessing tag quality. The methodology identifies three aspects of tag quality in a social tagging system: individual tags, collections of tags, and the association of the tags with the corresponding resources. The proposed methodology is being empirically evaluated using the tags of several social tagging systems. In addition to validating and refining the tag quality assessment methodology, the empirical analysis is expected to provide new findings about various properties of the tags. These findings will be used to improve the design of tagging systems by incorporating mechanisms that induce the contribution of high quality tags.