Cooling towers are critical engineered water systems for air conditioning and refrigeration but can create favorable conditions for Legionella pneumophila growth and aerosolization. Human exposure to L. pneumophila-contaminated aerosols can cause Legionnaire's disease. Routine monitoring of L. pneumophila in cooling towers offers possibilities to develop quantitative microbial risk assessment (QMRA) models to guide system design, operation, control, and maintenance. Here, we used the regulatory monitoring database from Quebec, Canada, to develop statistical models for predicting L. pneumophila concentration variability in cooling towers and integrate these models into a screening-level QMRA model to predict human health risks. Analysis of 105,463 monthly L. pneumophila test results revealed that the exceedance rate of the 104 colony-forming unit (CFU) per liter threshold was constant at 10 % from 2016 to 2020, emphasizing the need to better validate the efficacy of corrective measures following the threshold exceedances. Among 2852 cooling towers, 51.2 % reported no detections, 38.5 % had up to nine positives, and 10.2 % over ten. The gamma or the lognormal distributions adequately described site-specific variations in L. pneumophila concentrations, but parametric uncertainty was very high for the lognormal distribution. We showed that rigorous model comparison is essential to predict peak concentrations accurately. Using QMRA, we found that an average L. pneumophila concentration below 1.4 × 104 CFU L−1 should be maintained in cooling towers to meet a health-based target of 10−6 DALY/pers.-year for clinical severity infections. We identified 137 cooling towers at risk of exceeding this limit, primarily due to the observation or prediction of rare peak concentrations above 105 CFU L−1. Effective mitigation of those peaks is critical to controlling public health risks associated with L. pneumophila.