BACKGROUND: Indocyanine green fluorescence angiography (ICG-FA) enables real-time visualization of perfusion in reconstructive surgery. However, interpretation remains largely subjective and may be influenced by multiple factors. This study aimed to evaluate interobserver variability in ICG-FA interpretation and its ability to predict perfusion-related complications. METHODS: Consultants and residents from the Department of Plastic, Reconstructive and Hand Surgery at Amsterdam UMC assessed 30 ICG-FA videos of reconstructive cases, classifying perfusion as normal, suspicious, or compromised. Interobserver agreement was analyzed using Fleiss' Kappa. A generalized linear mixed model (GLMM) was used to evaluate the effect of experience level and the presence of a perfusion-related complication on assessment accuracy. RESULTS: Fifteen participants completed the survey (eight residents, seven consultants). Overall agreement was moderate (κ = 0.42), slightly higher among residents (κ = 0.44) than consultants (κ = 0.35). Observers were significantly less likely to correctly assess flap perfusion using ICG-FA in flaps with a perfusion-related complication compared with uncomplicated flaps (OR 0.04, 95% CI 0.003-0.41, p = 0.007). Observer experience (resident vs. consultant) did not significantly influence assessment accuracy (OR 1.08, 95% CI 0.52-2.23, p = 0.83). CONCLUSION: These findings demonstrate substantial interobserver variability in subjective ICG-FA interpretation and highlight the difficulty of visually identifying venous outflow disturbances and subtle perfusion abnormalities. Quantitative ICG-FA may help address these limitations.