A nearest neighbor representation of a Boolean function is a set of positive and negative prototypes in $R^n$ such that the function has value 1 on an input iff the closest prototype is positive. For $k$-nearest neighbor representation the majority classification of the $k$ closest prototypes is considered. The nearest neighbor complexity of a Boolean function is the minimal number of prototypes needed to represent the function. We give several bounds for this measure. Separations are given between the cases when prototypes can be real or are required to be Boolean. The complexity of parity is determined exactly. An exponential lower bound is given for mod 2 inner product, and a linear lower bound is given for its $k$-nearest neighbor complexity. The results are proven using connections to other models such as polynomial threshold functions over $\{1, 2\}$. We also discuss some of the many open problems arising.