Some studies have suggested that tail streamers in the barn swallow Hirundo rustica may have been elongated 10–12 mm by sexual selection, but according to other studies, the length of these feathers is at the aerodynamic optimum or very close to it. To shed light on this issue, outermost tail feathers were experimentally shortened in male and female barn swallows by 1, 11 or 21 mm. Changes in four physiological parameters commonly used to estimate phenotypic condition in birds (weight, erythrocyte sedimentation rate, blood leukocyte concentration and heterophil/lymphocyte ratio) were checked one month later. Health improved (blood leukocyte concentration decreased) in the group of birds with tails shortened by 11 mm (both males and females), but body condition deteriorated (weight decreased) compared to the other two experimental groups. There was no significant effect of tail-length manipulation on the other two physiological parameters. These contradictory results suggest trade-offs between components of phenotypic condition. Possible negative relationships between condition-related traits imply that using one or very few physiological parameters to estimate phenotypic condition might not be appropriate. The most plausible explanation for the turning point in phenotypic condition when streamers were shortened by 11 mm is that these feathers are 7–15 mm longer than the aerodynamic optimum in both sexes. Therefore, our results are consistent with the hypothesis that tail streamers have been elongated 10–12 mm by sexual selection. This conclusion disagrees with a previous study on the effect of experimental tail shortening on haematocrit, but the complexity of interpreting changes in haematocrit might account for this discrepancy.