It would be valuable to be able to infer the habitat quality of underwater-feeding birds and mammals from their diving or head-dipping behavior, especially when underwater sampling is difficult. In air-breathing animals that obtain their food by diving from the water surface to the bottom (or any specific depth), the underwater feeding time is expected to be a function of travel time and gain rate. Hence, for a given travel time, the underwater feeding time should reflect the (initial) quality of the patch where the animals forage. We applied this approach to estimate habitat quality of tundra swans digging for starch-rich belowground propagules of aquatic macrophytes on migratory stopover sites. Swans do not travel underwater but instead trample with their heads above water before dipping their heads, and the optimal dive model was adjusted for this. At an autumn staging site, habitat quality estimated from head-dipping time was not significantly different from locally measured habitat quality, but the uncertainty around the estimated quality was large. Significantly longer head-dipping time was recorded at a spring staging site, despite a shorter trampling time. This long underwater feeding time indicated that the habitat quality was lower, while the short trampling time was explained by more shallow burial of belowground propagules at the spring staging site than at the autumn staging site. Both features were corroborated with biomass data. It should be noted, however, that the precision of the indirect method of habitat assessment varies with habitat quality, being greater for low quality habitat, but probably still not very precise. Further research is needed to confirm the inferred low habitat quality of the spring staging site and to clarify how tundra swans complete their vernal migration.