The development of controlled-release formulations should be based on a clinico-pharmacological rationale such as increased compliance, reduced side effects and improved efficacy. The pharmacokinetic profile of a controlled-release formulation and its dose regimen should be compared under steady-state conditions with that of an immediate-release formulation or that of another controlled-release formulation. Apart from conventional characteristics such as AUC, tmax and Cmax, alternative characteristics are suggested such as residual concentration at the end of the dose interval, peak-through fluctuation in steady state, plateau time, statistical moments, in vivo input functions and intravenous infusion schemes which mimic the concentration/time profile after oral administration of the controlled-release formulation. The pharmacokinetic steady-state profile should be reproduced with and without food, from day to day, and at various dose levels. The in vitro specification should be based on in vivo requirements for "within-product bioequivalence".