Since its discovery, it has been generally assumed that the primary function of the zona glomerulosa of the adrenal cortex is the secretion of aldosterone. Taking evidence from the rat, and recognising that there is probably considerable species variation, I argue here that the glomerulosa in fact has many functions, including aldosterone synthesis, but is probably only a relatively poor de novo source of steroid. In vitro, the CYP11B2 (aldosterone synthase) of the glomerulosa can and does utilise as substrates products arising from CYP11B1 (11beta-hydroxylase) activity in fasciculata cells. Whether it does in vivo is open to question, but corticosterone and 18-hydroxydeoxycorticosterone are both present in circulating rat plasma at suitable concentrations. Such a mechanism would explain several inconsistencies in the literature, including the anomalous distribution of steroidogenic enzymes in the glomerulosa, the stimulation of CYP11B1 products by aldosterone secretagogues such as potassium ions or angiotensin II, the partial dependence of aldosterone secretion in vivo on an intact pituitary, the sensitivity of aldosterone secretion to tissue disruption in vitro, and the "late pathway" regulation of aldosterone synthesis.