There has been increased interest in the measurement of melatonin in plasma and saliva recently either as a marker of circadian phase or to understand the physiological role of melatonin. For both situations, there is a need for a specific assay for melatonin that is sensitive enough to detect low concentrations (<2 pg/mL). Since the mid‐1970s, there have been many assays developed to measure melatonin in blood and saliva. Radioimmunoassays and ELISA have predominated because of their relative simplicity and high throughput. In this review, I show that the early radioimmunoassays while providing valuable information about nocturnal melatonin levels in humans, generally produced inaccurate basal (daytime) levels. Mass spectrometry assays, however, have provided us with the target values that immunoassays need to achieve, that is, daytime plasma melatonin levels <1 pg/mL. There are now many contemporary commercial assays available utilising both RIA and ELISA technologies, but not all achieve the standards set by the mass spectrometry assays. The performance of these assays is reviewed. I conclude with recommendations on issues researchers need to consider when conducting melatonin studies, including the importance of time of day of collection, validation of assays, the potential causes of poor assay specificity at low levels, the advantages/disadvantages of using saliva vs plasma and extraction assays vs direct assays, kit manufacturers responsibilities and the reporting requirements when publishing melatonin studies.