Purpose: To evaluate the quality of pain assessment in Dutch ICUs and its room for improvement. Materials and methods: We used a modified RAND method to develop pain assessment indicators. We measured performance on the indicators using retrospectively collected pain measurement data from Dutch ICUs, which are all mixed medical – surgical, of three months within October 2016–May 2017. We assessed the room for improvement, feasibility of data collection, and reliability of the indicators. Results: We defined four pain assessment indicators. We analyzed 45,688 patient-shift observations from 15 ICUs. In 69.2% (IQR 58.7–84.9) of the patient-shifts pain was measured at least once (indicator 1); in 56.7% (IQR 49.6–73.5) pain scores were acceptable (indicator 2); in 11.7% (IQR 5.6–26.4) pain measurements with unacceptable scores were repeated within 1 h (indicator 3); and in 10.9% (IQR 5.1–20.1) unacceptable scores normalized within 1 h (indicator 4). We found data collection feasible because data were available for >79.3% of the admissions, and all indicators reliable as they produced consistent performance scores. Conclusions: There is substantial variation in pain assessment across Dutch ICUs, and ample room for improvement. With this study we took a first step towards quality assurance of pain assessment in Dutch ICUs.