Much of the empirical research in the field of eyewitness identifications focuses on factors affecting eyewitness accuracy. The current study focuses on when evidence-based practices for eyewitness identifications are used by police in the field. We specifically explored whether there are factors associated with which law enforcement agencies use best practices for lineups by analyzing secondary data from a large, national sample of law enforcement agencies (Police Executive Research Forum, 2013; N = 619). We focused on four evidence-based practices (double-blind lineups, witness instructions, documenting confidence, and increased training) and analyzed whether agency size and location were related to the use of these practices. We also specifically studied the role of training and its relationship to the use of best practices. Results showed agency size, but not location, was a significant factor with smaller agencies reporting using fewer best practices than larger agencies. Additionally, our results revealed consistent, strong associations between training and implementation of a variety of recommended identification procedures. Qualitative responses also confirmed the important role of training. Implications regarding law enforcement training about eyewitness identifications and the translation of eyewitness research into practice are discussed.