The early identification of talented individuals has become increasingly important across many performance domains. Current talent identification (TI) schemes in sport typically select on the basis of discrete, unidimensional measures at unstable periods in the athlete's development. In this article, the concept of talent is revised as a complex, dynamical system in which future behaviors emerge from an interaction of key performance determinants such as psychological behaviors, motor abilities, and physical characteristics. Key nonlinear dynamics concepts are related to TI approaches such as sensitivity to initial conditions, transitions, and exponential behavioral distributions. It is concluded that many TI models place an overemphasis on early identification rather than the development of potentially talented performers. A generic model of talent identification and development is proposed that addresses these issues and provides direction for future research.