Abstract Background Legionella species cause severe forms of pneumonia with high mortality and complication rates. Accurate clinical predictors to assess the likelihood of Legionella community-acquired pneumonia (CAP) in patients presenting to the emergency department are lacking. Methods We retrospectively compared clinical and laboratory data of 82 consecutive patients with Legionella CAP with 368 consecutive patients with non-Legionella CAP included in two studies at the same institution. Results In multivariate logistic regression analysis we identified six parameters, namely high body temperature (OR 1.67, p < 0.0001), absence of sputum production (OR 3.67, p < 0.0001), low serum sodium concentrations (OR 0.89, p = 0.011), high levels of lactate dehydrogenase (OR 1.003, p = 0.007) and C-reactive protein (OR 1.006, p < 0.0001) and low platelet counts (OR 0.991, p < 0.0001), as independent predictors of Legionella CAP. Using optimal cut off values of these six parameters, we calculated a diagnostic score for Legionella CAP. The median score was significantly higher in Legionella CAP as compared to patients without Legionella (4 (IQR 3–4) vs 2 (IQR 1–2), p < 0.0001) with a respective odds ratio of 3.34 (95%CI 2.57–4.33, p < 0.0001). Receiver operating characteristics showed a high diagnostic accuracy of this diagnostic score (AUC 0.86 (95%CI 0.81–0.90), which was better as compared to each parameter alone. Of the 191 patients (42%) with a score of 0 or 1 point, only 3% had Legionella pneumonia. Conversely, of the 73 patients (16%) with ≥4 points, 66% of patients had Legionella CAP. Conclusion Six clinical and laboratory parameters embedded in a simple diagnostic score accurately identified patients with Legionella CAP. If validated in future studies, this score might aid in the management of suspected Legionella CAP.