Despite the importance of safety climate assessment in healthcare, the psychometric properties of existing safety climate surveys have seldom been tested. We developed a German version of the HSOPSC and investigated its dimensionality and predictive validity. In adapting the survey we differentiated between the two management levels (hospital, unit) by duplicating dimensions of the HSOPSC. Based on a sample of n = 568, confirmatory and exploratory factor analyses (CFA, EFA) were applied. Relationships between safety climate and outcome variables were investigated by regression analyses. CFA did not replicate the proposed factor structure, supporting evidence from other European countries. EFAs revealed eight factors (instead of twelve proposed for the original instrument) and allowed for differentiating between management levels (hospital, unit). For physicians and nurses, different predictors were identified for overall safety climate perception. Further cross-cultural investigations of dimensionality are necessary for international comparisons of safety climate in healthcare. Our results highlight the importance of capturing the impact of different management levels, especially in large hospitals, and differences between professional groups in order to design interventions accordingly.