The evolutionary and behavioural ecology of migratory birds has received much theoretical and empirical attention. We contribute to this field by contrasting the weather at departure and stopover durations of a long‐distance migratory sandpiper prior to initiating lengthy transoceanic vs. transcontinental flights of potentially variable duration. Transoceanic flights provide few if any stopover options. We predicted that transoceanic migrants should therefore be more selective of energetically favourable weather at departure and have longer stopover durations prior to departing, using time as a surrogate for cumulative fuel acquisition, compared with transcontinental migrants. We used recent advances in capture–recapture modelling to quantify how weather conditions, length of stay, including estimated residence time prior to capture, and age class correlated with daily departure probabilities of Semipalmated Sandpipers Calidris pusilla at a coastal and an inland stopover site at comparable latitude. As expected, departure probabilities from both sites were higher with increasing strength of tailwinds, and the strength of this effect was larger for birds facing transoceanic vs. transcontinental flights. Cloud cover and temperature conditions at departure converged between sites at intermediate values from different background distributions. Stopover durations at both sites were substantially longer than needed if the birds were pursuing a simple tactic of arrive–fatten–leave at the stopover site. We infer that both sites provided high levels of both food and safety relative to other stages in the birds’ annual cycle, favouring lengthy stopovers and subsequent use of lengthy flights from both sites. Our study shows that recent advances of capture–recapture models can provide additional resolution to studies of the migration strategies of birds and refine our perspective on global patterns of migration routes and stopover decisions.

On their journey to wintering or breeding sites, migratory birds usually alternate between migratory flights and stopovers where they rest and refuel. Actually, migrating birds spend most of the time at stopovers. Consequently, selection to minimize total time spent on migration likely operates mainly on the effectiveness of stopover rest and refueling. Departure probability from stopover sites depends both on weather conditions and fuel stores, but their respective role has not been quantified. In the present study, we assess the relative contribution of factors driving the departure decision from a stopover site. As we cannot reliably characterize body condition and restness when capture probability is low, we propose to use the Time Since Arrival (TSA) as a proxy of the changes through days of the internal state of stopovering birds. We developed a specific capture-recapture model to quantify the relative contribution of TSA and climatic conditions on a 20-year capture-recapture dataset of a long-distance migratory songbird (Sedge warbler). The effect of TSA has yet the major contribution to departure probability compared to weather conditions. Here, low humidity and an increase of atmospheric pressure in the days preceding departure are the weather conditions associated with a higher departure probability but remain secondary compared to the time the individual has already spent at the site. The probability to depart from a stopover site is therefore largely determined by the time that a bird has already spent at the site. Whether this Time Since Arrival is rather a proxy of resting, feeding or fattening efficiency remains to be explored.