Birds on migration spend much more time on stopover sites to refuel for the next migration step than aloft, but empirical data on stopover duration are rare, especially for Palearctic trans-Sahara migrants whilst crossing the desert. Previous studies suggest that stopover duration of fat birds in oases is much shorter than that of lean birds. During 2003 and 2004 capture–recapture data of migrating passerines from two inland oases in spring and from one coastal site in autumn in Mauritania, West Africa, were analysed to test whether the probability of being a transient and the stopover duration depend on fuel stores at first capture. The application of capture–recapture models revealed that during autumn migration at the coast the proportion of transients (individuals that stop over only for 1 day) was relatively high (77–90%) in three out of four species investigated and stopover duration was short (1.9–4.6 days). In the inland oases in spring, transients were detected in only four out of 12 analyses. Stopover duration was longer than at the coast in autumn and surprisingly long in some species with durations of up to 30 days. Models taking into account the initial fat load of birds on the first capture occasion were, with one exception, never the most parsimonious ones. This indicates that the time spent after and before capture at the stopover site did not depend on the fat stores at first capture. Therefore, we cannot confirm the assumption that birds arriving at stopover sites in the desert with low fat loads stay longer than birds that arrive with high fat loads.