This study examined the relationship among resting cortisol levels measured in serum (SER), saliva (SAL). overnight urine (ON) and 24 h urinary free cortisol (24 h) samples to determine which would be most appropriate for monitoring recovery from the physiological stress imposed by exercise training. Eight (3 female, 5 male) physical education students (X+/-SD age= 22+/-2 y, mass= 73.6+/-17.1 kg; predicted 120m shuttle] VO2max = 3.83+/-1.34 L x min(-1); 51.4+/-8.0 mL x kg(-l) x min(-1)) volunteered. Venous blood, saliva and urine were collected following one day of recovery from training. Morning (0730-0800 h) cortisol was analysed commercially using solid phase radioimmunoassay. Mean resting cortisol concentrations for these samples were in the high end of the normal reference interval previously reported for this population in the literature: SER (656.5+/-122.3 nmol x L(-1)). SAL (41.9+/-12.3 nmol x L(-1)), ON (72.0+/-19.5 nmol x L(-1)) and 24 h (170.8 +/-58.7 nmol x day(-1)). Intraclass correlation coefficients (R) showed a high correlation between 24 h versus SER (R=0.99), SAL (R=0.97), and ON (R=0.97) cortisol, and between SER and SAL cortisol (r=0.99) Limits of agreement analysis to determine measurement error also revealed significant correlations (r2=0.60-0.99) between the dependent variables. In conclusion, the strong intraclass correlation coefficient between 24 h cortisol versus SAL and SER suggests that either sample can be used to monitor cortisol hormone during a recovery period from exercise training. The use of SAL samples for cortisol analysis may have many advantages due to its non-invasive nature.