The bio-availability of two slow-release theophylline capsules employing different sustained-release principles, was investigated in 12 children suffering from bronchial asthma, on this basis of the renal excretion of the unchanged active substance. With the aid of a model-bound convolution procedure, the completeness and the variability of absorption on multiple dosaging under non-steady-state conditions were established. Depending upon the retardation principle, both the mean bioavailability and its scatter differed from one preparation to the other, one of them being, on average, more completely absorbed than the other, while, at the same time, showing an appreciably more variable bio-availability. In consequence of the first pass metabolisation, the percentage of theophylline that was excreted unchanged in the urine, was greater with the intravenous route of administration than with peroral application. In the case of the secondary metabolite 3-methylxynthine, the situation was reversed.