We know relatively little about the evolutionary processes underlying endocrine physiology, despite the fact that hormones play a critical role in shaping an animal’s behaviour, ecology, and life history.  Comparative endocrine studies offer an exciting and unique method for examining the selective pressures that influence endocrine traits across species.  However, there are some unique considerations to be taken into account when comparing endocrine data across species. Failure to do so can lead to spurious results and biologically inappropriate conclusions.

Here, we identify the issues and assumptions that must be dealt with in comparative endocrine studies, particularly meta-analyses.  We outline the sources of error introduced by both biological and methodological factors, which are unique to endocrine data. In addition, we offer the first empirical demonstration of how these sources of error may affect the conclusions drawn from inter-specific comparisons. We conducted a cross-lab comparison of assay performance to examine how inter-lab variation contributes to variance in reported corticosterone values. Our results show that inter-lab variance in measured corticosterone levels is considerable, highlighting the strength of inter-lab effects when comparing data. Then, using birds as a case study because they represent a particularly well-studied taxonomic group, we compiled the largest corticosterone database to date.  We use this dataset to demonstrate that phylogenetic relationships can only be tested when effectively controlling for structured error in the database. Our empirical findings allow us to offer specific recommendations about how to control for sources of error and which metrics may yield the most biologically meaningful insights.  Comparative endocrinology is an exciting area of evolutionary physiology and has the potential to greatly enhance our understanding of how physiological traits evolve, but understanding the limitations of comparisons is important to allow the field to progress robustly.