Background: Nurses' clinical judgement plays a vital role in pressure ulcer risk assessment, but evidence is lacking which patient characteristics are important for nurses' perception of patients' risk exposure. Objectives: To explore which patient characteristics nurses employ when assessing pressure ulcer risk without use of a risk assessment scale. Design: Mixed methods design triangulating observational data from the control group of a quasi-experimental trial and data from semi-structured interviews with nurses. Setting: Two traumatological wards at a university hospital. Participants: Quantitative data: A consecutive sample of 106 patients matching the eligibility criteria (age ≥18 years, no pressure ulcers category ≥2 at admission and ≥5 days expected length of stay). Qualitative data: A purposive sample of 16 nurses. Methods: Quantitative data: Predictor variables for pressure ulcer risk were measured by study assistants at the bedside each second day. Concurrently, nurses documented their clinical judgement on patients' pressure ulcer risk by means of a 4-step global judgement scale. Bivariate correlations between predictor variables and nurses' risk estimates were established. Qualitative data: In interviews, nurses were asked to assess fictitious patients' pressure ulcer risk and to justify their risk estimates. Patient characteristics perceived as relevant for nurses' judements were thematically clustered. Triangulation: Firstly, predictors of nurses' risk estimates identified in bivariate analysis were cross-mapped with interview findings. Secondly, three models to predict nurses' risk estimates underwent multiple linear regression analysis. Results: Nurses consider multiple patient characteristics for pressure ulcer risk assessment, but regard some conditions more important than others. Triangulation showed that these are measures reflecting patients' exposure to pressure or overall care dependency. Qualitative data furthermore indicate that nurses are likely to trade off risk-enhancing conditions against conditions perceived to be protective. Here, patients' mental capabilities like willingness to engage in one owns care seem to be particularly important. Due to missing information on these variables in the quantitative data, they could not be incorporated into triangulation. Conclusions: Nurses' clinical judgement draws on well-known aetiological factors, and tends to expand conditions covered by risk assessment scales. Patients' care dependency and self-care abilities seem to be core concepts for nurses' risk assessment.