A model of intensive care nursing of patients and their significant others compatible with a data system was developed and evaluated by using an action research approach in 1997–2001. The study was carried out in Oulu University Hospital’s ICUs for emergency and postoperative care and internal medicine. At stage I of the study, the model of intensive care nursing was developed and pilot-tested by collecting questionnaire data from the nurses who participated in the action research group (N = 14), from the texts produced by small groups made out of the action research group and the whole group (N = 33), by audio-recording oral report sessions (N = 57) and by using the researcher’s written notes. The data for pilot testing were collected on a form from nursing documents (N = 17) and on a questionnaire from nurses (N = 11). The research data were analysed with methods of inductive and deductive content analysis. At stage II, the model was made compatible with the ICU data system. At stage III, the validity of the model was evaluated by analysing patient data drawn from the emergency and postoperative ICU’s data system (N = 1464) and patient records filed by the researcher (N = 30). The patients’ needs for nursing and the quantitative needs for staff were compared between the model of intensive care nursing and the Therapeutic Intervention Scoring System (TISS) based on the patient data drawn from the data systems of the emergency (N = 253) and postoperative (N = 579) ICU. The data were analysed using statistical methods and methods of content analysis. The model of intensive care nursing includes nursing diagnoses of the changes in the patient’s vital functions, the restrictions and experiences caused by the disease and its treatment and the significant others’ distress, the nursing to meet the severity of the health problems, the nursing interventions and the outcomes of nursing as well as the nursing workload. Based on the validity assessment, the model differentiated between the nursing interventions needed by the patients in different admission types and with different severity scores (p