Evolving SARS-CoV-2 variants and changing levels of pre-existing immunity require re-evaluation of antigen-detecting rapid diagnostic test (Ag-RDT) performance. We investigated possible associations between Ag-RDT sensitivity and various potential influencing factors, such as immunisation status and viral variant, in symptomatic hospital employees. In this observational study, RT-PCR, Ag-RDT, and symptom-specific data were collected at three SARS-CoV-2 test centres for employees of the Charité-Universitätsmedizin Berlin hospital (Berlin, Germany). Employees reporting SARS-CoV-2-like symptoms, those at an increased risk of infection (eg, due to contact with an infected person), those testing positive in a previous self-administered Ag-RDT, or those seeking release-testing to return to work at least 7 days after a positive RT-PCR test were eligible for combined testing by RT-PCR and Ag-RDT. Only data from individuals with an ongoing SARS-CoV-2 infection as assessed by RT-PCR were used for further analysis. Bayesian regression analyses were done to evaluate possible differences in viral load and Ag-RDT sensitivity according to viral variant and immunisation status (previous vaccination or recovery from infection), using data from first RT-PCR positive samples in an infection. A comprehensive logistic regression analysis was used to investigate potential concomitant associations between Ag-RDT sensitivity and level of pre-existing immunity, time post symptom onset, viral load, gender, age, and Ag-RDT device. Ag-RDT performance was also compared between supernatants from cell cultures infected with the omicron variant of concern (VOC) or the wild-type strain (pre-VOC). Between Nov 30, 2020 and Feb 11, 2022, a total of 14 773 samples from 7675 employees were tested for SARS-CoV-2 by both RT-PCR and Ag-RDT. We found a negative association between immunisation status and Ag-RDT sensitivity in symptomatic employees, with an observed sensitivity of 82% (94% highest posterior density interval [HPDI] 78-86) in immunologically naive participants compared with 73% (68-78) in multiply immunised individuals (ie, those with at least two vaccinations or recoveries from infection) and median log10 viral loads of 7·02 (IQR 5·83-8·07) and 8·08 (6·80-8·89), respectively. The dominant viral variant changed several times during the study period, from the pre-VOC period (sensitivity 80% [94% HPDI 75-85] in symptomatic participants) through the alpha variant (82% [70-94]), delta variant (75% [69-82]), and omicron variant (72% [65-79]) waves, concomitantly with a steep increase in vaccination coverage in our dataset. In a comparison of Ag-RDT performance on cell culture supernatants, we found no difference between the wild-type and omicron viral variants. On the basis of our findings and data from other studies, we hypothesise that the observed reduction in clinical Ag-RDT sensitivity, despite higher SARS-CoV-2 RNA loads, is due to shorter incubation times later in our study period resulting from increased population immunity or changes in immune response dynamics caused by later SARS-CoV-2 VOCs. Berlin University Alliance, German Ministry of Education and Research, the EU (Projects EU4Health and ReCoVer), and the Berlin Institute of Health.