H ormones influence behaviour, and are also influ-enced by behaviour. Monitoring their levels can therefore provide insights into the mechanistic aspects of behaviour. In male mammals for example, elevated levels of testosterone are associated with increased aggression and dominance (Creel et al. 1993, 1997; Mazur & Booth 1998) and in social mammals, levels of stress hormones (e.g. corticosterone, glucocorticoid and cortisol) are associated with rank (Sapolsky 1985; Creel et al. 1996, 1997). Research has associated hormone levels with dif-ferent behaviours such as sexual, reproductive, courtship, parental, aggressive and feeding behaviours. Comparative tools for hormonal analysis provide insights into evolu-tionary theories based on behavioural aspects, such as reproductive suppression and the 'challenge hypothesis' (e.g. Creel et al. 1993). In field studies, hormones are usually extracted from blood samples, or noninvasively from saliva, urine and faeces (Creel et al. 1992; Cavigelli 1999; Hirschenhauser et al. 1999; von Engelhardt et al. 2000). Samples derived from trapped or handled animals are problematical because stress may alter blood and urine hormonal levels (Creel et al. 1992). Additional problems with blood samples are that they are not always available, the amount that can be taken at a given time is limited, and various safety and ethical issues exist. Furthermore, blood and saliva must be transported cold or frozen, conditions that are sometimes difficult to obtain in the field (Yang et al. 1998). Urine and faeces samples are sometimes diffi-cult to obtain from free-ranging animals that cannot be continuously observed, or from species that deposit in common latrines. An alternative source for hormones may be found in hair, which can be collected noninvasively, and is already used to extract DNA (Woodruff 1993; Morin et al. 1994), trace metals, naturally occurring compounds and drugs (Wheeler et al. 1998). Hair is safe, readily available, and easy to store and transport. Hair sampling does not involve pain or possible infection, and the analysis is unaffected by the momentary stress of capture (Yang et al. 1998). Hair analysis may allow one to monitor hormonal changes over weeks or months (between moults; Maurel et al. 1986) by shaving off a patch of hair and resampling the newly grown hair. Hormonal hair analysis offers only a long-term profile, however, and is not suitable for monitoring hourly or daily (short-term) fluctuations in hormonal levels. It provides the resolution needed for studies of main behavioural trends, especially in stable hierarchical social systems. Hair has already been used to diagnose early pregnancy in cows by detection of proges-terone (Liu et al. 1988), to detect oestradiol and testoster-one in cattle (Gleixner & Meyer 1997) and anabolic steroid and corticosteroid abuse in athletes (Bowers & Segura 1996; Hold et al. 1999; Kintz et al. 1999; Cirimele et al. 2000). In humans, the levels of steroid hormones in hair do not vary significantly between different regions of the scalp (Wheeler et al. 1998). Oestradiol, progesterone and testosterone levels measured in healthy human adults' hair correlate significantly with the levels measured in their serum (Yang et al. 1998). As an example of the utility of this method, we use data from our long-term study on rock hyrax, Procavia capensis. Observations of urination are rare in this species, and shared defecation sites do not allow individual identification of faeces. Extracting hormones from hair samples taken from live captured rock hyrax is an easy method, which enables reliable monitoring of long-term trends in hormonal changes despite the stress caused by trapping (Koren 2000).