Over the course of 1 year, a patient with a rapid cycling bipolar affective disorder was followed at weekly intervals to examine whether plasma hormones and urinary catecholamines could predict current or future mood. Higher cortisol levels were found to predict depressed mood 3 days after blood sampling, higher urinary dopamine predicted a manic mood 3 days after blood sampling, urinary norepinephrine was associated with severity of current mood and prolactin was lower with concurrent depressed mood. In multivariate analyses of mood against cortisol, prolactin and three urinary catecholamines, > 50% of the variance in mood state in 3 days was explained by combinations of these biologic measures, especially cortisol and urinary dopamine, while all five biologic variables contributed to explaining 50% of the variance in current mood state. Based on the interrelationships between urinary dopamine, norepinephrine and mood, we postulate the existence of an overcompensating mechanism which is reflected in opposing correlations between urinary dopamine and norepinephrine with mood, despite the two urinary catecholamines being positively correlated.