Paper considers the case when end point in clinical trials for comparison of drug effect is measured at several time intervals and its time course is clinically significant. Authors experimented and propose methods for comparison of curves in biopharmacy to be used in analysis of clinical curves, instead of usual comparison of mean curves. It were considered three types of comparisons from biopharmacy: comparison between dissolution curves (dissolution metrics), comparisons between plasma levels (bioequivalence metrics) and comparison between dissolution kinetics and pharmacokinetics of modified release drugs, so-called in vitro- in vivo correlations (IVIVC). Particular application considered pain curves in studies comparing analgesic effect of drugs and comparison of effects of antiaggregant drugs on erythrocytes sedimentation curves. Mathematically methods proved to be easy applicable but in all particular cases the validation of methods was connected with defining of thresholds beyond which “distances†are clinically significant. It concerns IVIVC, in case of pain curves comparison was practically an in vivo – in vivo comparison between plasma levels of an active component and time course of clinical effect. Since plasma levels were instantaneous values and pain curves represented practically a cumulative effect, direct comparison was not possible. After replacing of pain curve with “differential pain curveâ€, i.e. the difference of pain intensity between two consecutive measuring points, the curves had practically the same “absorption – distribution†plasma level profile and correlation became meaningful. It was practically a successful pharmacokinetic- pharmacodynamics (PK/PD) modelling. It has been argued that this type of methods are particular cases of comparison between input – output correlation in the “black-box†model with the application of Laplace transform in estimating “transfer function†in the general dynamic systems theory. Consequently, the methods can be extrapolated farther to comparison of more large classes of time dependent endpoints.