Objectives: Capillary refill time (CRT) has been taught as a rapid indicator of circulatory status but to be a useful
clinical test, CRT needs to be reproducible when performed by another health care worker. No inter-rater agreement studies have been reported for adult patients. The aim of this study was to determine the inter-observer reliability of CRT in a sample of adult emergency department (ED) patients. Methods: This prospective observational study included clinically stable ED patients with a variety of conditions from two community EDs. A doctor and a nurse each measured CRT by estimation to the nearest half-second using a standard method on each patient. They were blinded to each other's measurements. The primary outcome of interest was inter-rater agreement. Secondary outcome was agreement in classification as normal or abnormal according to accepted definitions. Data was analysed using bias plot analysis, correlation, absolute percent agreement and kappa analysis. Results: Totally, 209 patients were enrolled; 51% were female and 86% were Caucasian. Median CRT was 2 seconds
(95% CI 2-2.35 seconds). The mean difference between measurements by the different observers was 0 second, 
however the 95% limits of agreement were very wide (-1.7 to +1.9 seconds). Agreement was 70% for classification of 'normal' or 'abnormal' using the 2-second definition of normal, with a kappa of 0.38. Conclusion: Interobserver
agreement in measurement of CRT was poor in adult subjects with wide limits of agreement. This is a serious threat to the appropriateness of this test for use in clinical practice.