Distinguishing among biological, pharmacological, and behavioural variability is essential for proper interpretation of the therapeutic experiment at each return visit. Within the behavioural component, partial compliance refers to all suboptimal levels of concordance between the patient's behaviour and the clinical prescription. However, the assessment of compliance is limited largely by imperfect measures of medication taking which are frequently distant in time and space from the medication-taking event itself. Most studies indicate compliance levels of only 50-70% with antihypertensive regimens as well as considerable variance from day to day and person to person. Therapeutic outcome may be a misleading method to assess the sufficiency of a regimen because of the high prevalence of suboptimal medication-taking behaviour. In selected situations, prolongation of pharmacological effect may compensate for imperfect medication-taking behaviour, confirmed by electronic medication monitors. Such pharmacodynamic prolongation exemplifies therapeutic sufficiency, a new paradigm for therapeutics in the 1990s.