Background In 2021, the Norwegian Institute of Public Health established temporary registry-based surveillance of severe acute respiratory infections (SARI). We aimed to describe the surveillance system and evaluate selected attributes to inform the establishment of a permanent SARI surveillance system. Methods SARI cases were defined using ICD-10 discharge codes from national health and administrative registries, including codes for acute upper or lower respiratory infection (URI, LRI), COVID-19, acute respiratory distress syndrome, pertussis, or otitis media. Data from polymerase chain reaction (PCR) analyses were available for 10 respiratory pathogens including SARS-CoV-2, influenza virus, and respiratory syncytial virus (RSV). We included data from 28.9.2020–31.3.2024 and calculated the following parameters: the proportion of cases tested for SARS-CoV-2, influenza virus and/or RSV; time between admission and registration of a SARI-related ICD-10 code; and proportion of cases with URI, LRI and COVID-19. Results We identified 214,730 SARI cases, of whom 82%, 73% and 53% were tested for SARS-CoV-2, influenza virus and RSV. Case peaks were predominantly driven by one or a combination of these pathogens. Median time between admission and a registered SARI diagnostic code was 5 (lower-upper quartile 3-10) days. Nowcasting and alternative case definitions for SARI with COVID-19, -influenza, and RSV improved the timeliness. The ICD-10 codes for LRIs and COVID-19 captured only ~55% of the cases in the age group 0-29 years compared to the routine case definition, where URIs were included. Conclusions Registry-based SARI surveillance provides timely data for handling epidemics of respiratory infections in Norway. We recommend establishing a permanent SARI surveillance system.