Supplementing local police forces is a burgeoning multibillion-dollar private security industry. Millions of formal surveillance agents in public settings are tasked to act as preventative guardians, as their high visibility presence is hypothesized to create a deterrent threat to potential offenders. Yet, rigorous evidence is lacking. We randomly assigned all train stations in the South West of England that experienced crime into treatment and controls conditions over a six-month period. Treatment consisted of directed patrol by uniformed, unarmed security agents. Hand-held trackers on every agent yielded precise measurements of all patrol time in the stations. Count-based regression models, estimated marginal means and odds-ratios are used to assess the effect of these patrols on crimes reported to the police by victims, as well as new crimes detected by police officers. Outcomes are measured at both specified target locations to which security guards were instructed to attend, as well as at the entire station complexes. Analyses show that 41% more patrol visits and 29% more minutes spent by security agents at treatment compared to control stations led to a significant 16% reduction in victim-generated crimes at the entirety of the stations' complexes, with a 49% increase in police-generated detections at the target locations. The findings illustrate the efficacy of private policing for crime prevention theory.