The scientific literature on tricyclic antidepressants contains few studies on the dose-effect relationship, and this aspect of the treatment has for many years been given little attention. However, recent reviews point out inadequate dosing as an important factor in "drug resistant depressions." Tricyclic antidepressants have a narrow therapeutic range and standard doses, which are frequently recommended, will not be therapeutic in some patients and toxic in other patients. Use of flexible dose schemes, which has been customary in clinical trials, carry a considerable risk of underdosing because common side effects occur at subtherapeutic doses. Dosing according to therapeutic effect is not feasible in light of the slow and variable rate of response. Much of the difficulties in establishing a dose-effect relationship appears to be related to the considerable pharmacokinetic variability. The major source of this variability is the genetic polymorphism related to the sparteine/debrisoquine oxygenase. For some of the tricyclic antidepressants, a concentration effect relationship has been established, but almost exclusively on the basis of retrospective studies. The dose-response problem is particularly important in clinical trials. Apparent differences or equivalence between a new drug and the control therapy thus may entirely be related to differences in doses of the two drugs. For tricyclic antidepressants, the use of flexible dose schedules and poor control of compliance may often lead to underdosing and a response rate below the real potentials of these drugs. Underdosing combined with a high rate of placebo response will increase the type 2 error risk considerably and may ultimately lead to the introduction of drugs that are less effective than the classical drugs.