Introduction Sources of infection of most cases of community-acquired Legionnaires’ disease (CALD) are unknown. Objective Identification of sources of infection of CALD. Setting Berlin; December 2016–May 2019. Participants Adult cases of CALD reported to district health authorities and consenting to the study; age and hospital matched controls. Main outcome measure Percentage of cases of CALD with attributed source of infection. Methods Analysis of secondary patient samples for monoclonal antibody (MAb) type (and sequence type); questionnaire-based interviews, analysis of standard household water samples for Legionella concentration followed by MAb (and sequence) typing of Legionella pneumophila serogroup 1 (Lp1) isolates; among cases taking of additional water samples to identify the infectious source as appropriate; recruitment of control persons for comparison of exposure history and Legionella in standard household water samples. For each case an appraisal matrix was filled in to attribute any of three source types (external (non-residence) source, residential non-drinking water (RnDW) source (not directly from drinking water outlet), residential drinking water (RDW) as source) using three evidence types (microbiological results, cluster evidence, analytical-comparative evidence (using added information from controls)). Results Inclusion of 111 study cases and 202 controls. Median age of cases was 67 years (range 25–93 years), 74 (67%) were male. Among 65 patients with urine typable for MAb type we found a MAb 3/1-positive strain in all of them. Compared to controls being a case was not associated with a higher Legionella concentration in standard household water samples, however, the presence of a MAb 3/1-positive strain was significantly associated (odds ratio (OR) = 4.9, 95% confidence interval (CI) 1.7 to 11). Thus, a source was attributed by microbiological evidence if it contained a MAb 3/1-positive strain. A source was attributed by cluster evidence if at least two cases were exposed to the same source. Statistically significant general source types were attributed by calculating the population attributable risk (analytical-comparative evidence). We identified an external source in 16 (14%) cases, and RDW as source in 28 (25%). Wearing inadequately disinfected dentures was the only RnDW source significantly associated with cases (OR = 3.2, 95% CI 1.3 to 7.8) and led to an additional 8% of cases with source attribution, for a total of 48% of cases attributed. Conclusion Using the appraisal matrix we attributed almost half of all cases of CALD to an infectious source, predominantly RDW. Risk for LD seems to be conferred primarily by the type of Legionella rather than the amount. Dentures as a new infectious source needs further, in particular, integrated microbiological, molecular and epidemiological confirmation.