Information retrieval systems may often deliver biased or unfair results, and before one can mitigate such unfairness, one must first measure it. In this chapter, we consider a number of information retrieval methods, and discuss metrics to measure the fairness of the output from such methods. We begin with a discussion of word embeddings, which represent terms as numerical vectors and are often used in downstream applications. Many popular word embeddings are known to exhibit undesirable bias, including gender bias. We then discuss search result ranking, in which bias may result in certain demographic groups being overrepresented or underrepresented.