This online supplement contains datasets (raw data) and study material collected in an experimental study by the University of Münster, Germany. The study is part of a larger research project (https://www.brawa.ovgu.de/en/) and examined the perception of acoustic fire alarm signals. Hazards like fires occur regularly and can cost people’s lives. Optimal auditory alarm signals enable laypeople to recognize dangers and to protect themselves. Existing fire alarm sound research focuses on alarm sounds and voice alerts presented singularly. We explored a combination of both and aimed to identify alarm signals that work optimally in everyday life. Thus, we conducted two online experiments: In Study 1 (N = 379), we tested eight alarm sounds regarding their typicality, their familiarity, their arousal, their valence, and their dominance. Siren-like alarm sounds were most effective. In Study 2 (N = 206), we combined the four most effective alarm sounds with a voice alert. The voice alert reinforced ambiguity reduction, action motivation, and action intention. Hence, we suggest using alarm sounds with siren-like patterns. They should be combined with a voice alert to foster a quick and specific (target task-oriented) reaction. The ethics committee of the University of Münster approved the studies (ID 2021-57-MT), and we preregistered both studies with AsPredicted.org under numbers #77031 and #81137 (see https://aspredicted.org/vx2rr.pdf and https://aspredicted.org/mt9g3.pdf). The studies were supported by the German Federal Ministry of Education and Research (grant numbers 13N15416 and 13N15419). This online supplement includes: Two codebooks describing all instructions and items in Study 1 and in Study 2 Raw data (anonymized) and analysis scripts (Note: The raw data contains only the information of persons who were included in the analysis and who gave their informed consent. Some demographic information was deleted to ensure anonymity.) Study material: Alarm signal example Hearing test implemented in both studies