Fat is the prime energy source for birds during prolonged exercise, but protein is also catabolized. Estimates of the amount of catabolizable fat and protein (termed fat and protein fuel) are therefore important for studying energetics of birds. As fat and protein fuel can only be measured by sacrificing individuals or by use of technically complex methods, scoring systems were invented to estimate fat and protein fuel of birds in the field. The visible subcutaneous fat deposits and the thickness of the flight muscles are each scored on an ordinal scale but these scales do not correspond linearly to fat and protein fuel within species, which is needed for analyses such as flight range estimates. We developed an anova-type model to estimate fat and protein fuel from fat scores (FS) and muscle scores (MS) along with total mass and a size measurement. Using data from 11 337 individuals of eight passerine species (Common Nightingale Luscinia megarhynchos, Eurasian Reed Warbler Acrocephalus scirpaceus, Melodious Warbler Hippolais polyglotta, Willow Warbler Phylloscopus trochilus, Orphean Warbler Sylvia hortensis, Garden Warbler Sylvia borin, Common Whitethroat Sylvia communis, Subalpine Warbler Sylvia cantillans) mist-netted in Mauritania, West Africa, we tested for independence of FS and MS and for variation in the relationship between scores and associated mass in response to physiological state. FS, MS and third primary length (size) explained variation in body mass of all eight species analysed (R2: 0.56–0.77). The parameter estimates of the model showed that fat and protein fuel increased monotonically with increasing fat and muscle scores. In two species we found small differences in the estimates between physiological states (seasons). We evaluated our model by comparing the predicted body mass of birds with both FS and MS equal to 0 with the mean body mass of individuals mist-netted with both scores equal to zero. The values were very close. The amount of fat extracted from dead Garden and Willow Warblers was within the range of predicted fat fuel derived from the model. We conclude that our model is a useful non-invasive method to estimate simultaneously mean fat and protein fuel of small passerines and we provide recommendations on its use.