The human iris is extremely unique, stable, and externally visible; making it ideally suited as a biometric modality. The performance of state-of-the-art in iris biometrics is approaching perfect correct recognition rates in small deployments with controlled image acquisition. Iris recognition systems need to improve continually in order to keep pace with the increasing number of users enrolled in applications, and in order to perform well in less controlled image acquisition circumstances. In this work we present improvements to all four stages of the biometrics pipeline that increase the accuracy of iris biometrics systems. The effects of mascara and other topical cosmetics on iris recognition are evaluated. A method for detection of soft contact lenses using modified binary patterns is introduced and is shown to accurately detect the presence of a clear, soft contact lens, while not reducing the correct detection rate of textured lenses. This work is further extended by using Binarized Statistical Information Feature (BSIF) in the place of binary patterns. BSIF is shown to outperform the binary pattern algorithm for this use case. The impacts of novel lens types and novel sensors are both also evaluated. A set of improvements to the iris biometrics gallery using multiple samples and regression analysis of quality metrics are also proposed and evaluated. These methods are shown to improve the correct identification rate of an iris biometrics system using real-world scenarios.