A method is described, intended for measuring the rate--over a short period of time--of aqueous humour outflow in non-anaesthetized primates. [125I] and [131I]-o-iodohippurates were infused s.c. at an interval into the conscious Cynomolgus monkey so as to give two time-displaced plasma curves. The isotope concentrations were measured in plasma by frequent sampling over the first 1-2 hr and in aqueous humour at the end of the experiment under brief anaesthesia. The coefficient of diffusion (kdiff) between plasma and aqueous, and the coefficient of loss (kout) out of the anterior chamber were calculated by making the usual assumptions about aqueous dynamics. Topical treatment of one eye with timolol consistently reduced kout, as expected, but many control values for kout were lower than expected or (impossibly) negative. Examination of the data reveals a likely source of error emanating from the commonly made assumption that diffusion of marker substance (in our case iodohippurate) from plasma to aqueous humour is proportional to the instantaneous concentration difference between plasma and aqueous. In fact, there may be a delay of several minutes during which the substance negotiates the stroma and anterior boundary layer of the iris. Evidence is presented that variation between and within individual monkeys may stem from the highly variable anatomy of the iris and from changes in sympathetic tone operating on the blood vessels and perhaps the anterior boundary layer.