Sentinel species are biological monitors that accumulate a pollutant in their tissues without significant adverse effects. Primarily used to measure the amount of a pollutant that is biologically available, they may also increase the sensitivity of an analytical procedure or summarise a complex pollution signal. This payer examines the validation of sentinels, referring particularly to the use of invertebrates in monitoring toxic metal pollution in terrestrial ecosystems. Few studies measure the capacity of a sentinel to quantify changes in ambient levels. Definitions of 'bioavailability' also differ between workers and few note that this: may be a highly species-specific measure. Validation requires that the temporal and spatial scales over which a sentinel integrates a pollution signal are known. The sentinel has to be calibrated against source concentrations and this relationship shown to be consistent over the normal range of exposure. This requires some consideration of the environmental and biological determinants of pollutant assimilation. Differences between populations can confound simple comparisons between sites based on native populations. Transplanted individuals, matched for age, sex and physiological state, might be used when the aim is not to assess bioavailability to the resident population. A simple measure of their capacity to detect differences in ambient pollutant levels is proposed to evaluate candidate species, to assess their consistency and capacity to equilibrate with their source. A small survey of earthworms from a well-defined gradient of Pb pollution is used to illustrate problems of interpreting tissue concentrations in sentinels. (C) 2001 Elsevier Science Ltd. All rights reserved.