Research using biometric data relies on consistent measurements within, and often among, observers. However, research into the relative importance of intra- and inter-observer variability is limited. More importantly, the influence of biometric variability on accurate quantification of biometric-based condition indices has not been analysed: it is unclear whether multiple errors become magnified or cancel one another out. Here, we quantify intra- and inter-specific variability in multiple biometrics, and derived condition indices, using museum bird specimens. Inter-observer variability was higher than intra-observer variability for all parameters. Measurement error (ME) varied from < 1% to > 50% for different biometrics. ME was magnified in condition estimates, reaching > 80% within-observers and > 90% among-observers. Significant differences in mean measurements were found for 17% and 67% of biometrics within-and among-observers, respectively; for condition indices, the figures were 50% and 67%, respectively. We discuss the implications of these findings for research into species' ecology, taxonomy and behaviour.