Albumin has multiple functions and is used in the clinical assessment of liver function, kidney function and nutritional status. However, few epidemiological studies have evaluated the association between sleep duration and albumin. Therefore, we carried out a cross-sectional study to address this issue. The aim of the study was to investigate the association between sleep duration and albumin in American adults based on the NHANES (National Health and Nutrition Examination Survey). A total of 9,973 participants aged $$\ge$$ 20 years were included in this study from NHANES 2015–2018. Weighted data were calculated according to analytical guidelines. Linear regression models and smooth curve fitting were used to assess and describe the relationship between sleep duration and albumin. The inflection point was determined by a two-step recursive method. Moreover, univariate and stratified analyses were performed. There was an inverted U-shaped association between sleep duration and albumin levels. Albumin levels were highest when the sleep duration was 7.5 h. Compared to 7–8 h of sleep, short sleep duration was linked to lower albumin levels [sleep duration $$\le$$ 5 h: β $$=$$ -1.00, 95% CI (-1.26, -0.74), P < 0.0001]. Compared to 7–8 h of sleep, long sleep duration was related to lower albumin levels [sleep duration $$>$$ 9 h: β $$=$$ -0.48, 95% CI (-0.68, -0.27), P < 0.0001]. Sleep duration had an inverted U-shaped relationship with albumin, with short or long sleep duration associated with significantly lower albumin levels.