The growing list of major police innovations demonstrates the willingness of many police administrators to take risks, incur costs, and devote energy to the process of inducing change. Analytical research can make administrators sensitive to the probable existence of an important relationship or uncover a pattern among cities, precincts, or officers. Most studies are cross-sectional (they measure a phenomena in several cities at a single time) and employ change in one variable is associated with a given change in another variable, holding all other variables constant. The results of such studies seem to support the proposition that the higher the probability of an arrest for a given reported crime, the lower the rate at which the crime occurs. Futhermore, an association between the probability of an arrest and police use of certain aggressive patrol strategies is suggested in other studies. Experimentation is used to confirm or reject the implications of analytical research. Effective experimentation calls for independent evaluation, clearly stated measures of valued outcomes, experimental control, and a set of reliable facts for comparisons. In addition, the understanding and cooperation of participating personnel must be assured. Recent experiments have demonstrated that police motor patrol is not the most effective form of patrol; that effective policing involves public participation; that a single, most effective police strategy does not exist; and that officers appear to differ significantly in their ability or willingness to make arrests. The outpouring of research and experimentation has provided the country with a large pool of research talent and with a great number of police administrators who have successfully applied analytical and experimental modes of thought to the management of their departments. Notes and references are included.