A cross-sectional study of 160 water samples collected from 72 cooling towers in 4 hospitals, 7 department stores, and 3 hotels in Bangkok was carried out to investigate Legionella pneumophila contamination and its predictive factors. All water samples were cultured for Legionella spp. and tested for L. pneumophila by real-time polymerase chain reaction (PCR). Some cooling tower parameters were measured and recorded. Data were analyzed using χ 2 -test, odds ratio and stepwise logistic regression analysis at the significant level of α = 0.05. Results revealed that the Legionella spp. contamination was 20.0% (32/160) and for L. pneumophila was 61.3% (98/160). The sensitivity of real-time PCR was higher than that of the culture. Factors significantly associated with L. pneumophila contamination by χ 2 -test were: the cooling tower model, size, use duration, pH, water temperature, use of ozone, and residual free chlorine (95% CI of OR > 1.0, P < 0.05). After stepwise logistic regression analysis, four predictive factors remained. These included the cooling tower model being a cross-flow type (adjusted OR = 3.1, 95% CI = 1.2-7.8, P = 0.017), use duration >5 years (adjusted OR = 3.6, 95% CI = 1.3-10.1, P = 0.016), water temperature <29.4°C (adjusted OR = 7.9, 95% CI = 2.1-29.6, P = 0.002), and residual free chlorine <0.2 ppm (adjusted OR = 8.5, 95% CI = 2.1-34.9, P = 0.003). Additionally, the risk probability for L. pneumophila contamination was estimated to be 13.9-97.1%, depending on the combination of predictive factors.