The tricyclic SSRI antidepressant association is often used in the treatment of resistant depressive illness. The pharmacokinetic interaction existing between these two types of drugs is well known, with as result, an increase of tricyclic antidepressant plasma levels. The aim of this work was to assess the clinical tolerance of the association of fluoxetine and tricyclic antidepressants, prescribed at usual doses. In 10 patients, having a bad response to a tricyclic antidepressant treatment, with in the therapeutic window adjusted plasma levels since 3 weeks, an association of fluoxetine (20 mg/d) to the tricyclic was prescribed. The other associated treatments were unmodified. The clinical evolution was recorded with the MADRS and the UKU scale for side effects, before the tricyclic antidepressant treatment adjustment (D-21) and just before the fluoxetine association (D1) and every 7 days after this association too. The tricyclic plasma levels (amitriptyline and clomipramine) and the patient phenotype CYP 2D6 and 2C19 were determined before and 7 days after the fluoxetine addition. A good clinical evolution was noted since the 7th day after the fluoxetine association to tricyclic (mean MADRS scores on D-21, D1, D7 and D14; 35.4, 33.1, 23.9, 16.8 respectively). In 3 patients, an anxiety increase on day 6, 14 and 16 respectively, after fluoxetine addition, induces a stop of the serotonergic antidepressant. In one patient all the treatment was stopped due to the appearance of a mood inversion. In another patient, after 14 days of antidepressant association, EC were prescribed as asked by the patient, due to an insufficient mood improvement, with a good clinical result and tolerance. The evolution of the side effects was surprising. There was no increase of the UKU score mean during the associated treatment, despite an increase of the tricyclic plasma levels that reached, in three patients, the toxic level (510, 605 and 860 ng/ml of amitriptyline + nortriptyline or clomipramine + demethylclomipramine). The UKU psychic score mean significatively decreased (7.7, 6.8, 5.3, 4 on D-21, D1, D7, D14 respectively). The fluoxetine association did not modify the neurological, neuro-endocrinologic and the skin side effects. None increase of headheck was noted. The increase of anxiety, observed in 3 patients, was not considered as a side effect of the antidepressant association, but an effect of the stimulant potency of fluoxetine in anxious patients. The pharmacogenetic results confirmed the strong inhibition potenty of fluoxetine on the CYP 2D6 isoenzyme. In 5 patients indeed, the extensive metabolizer phenotype was modified in a poor metabolizer phenotype, seven days after the association of fluoxetine. The CYP 2C19 phenotype was unchanged after this association. The patient phenotype did not seem to interfere with the clinical results. In conclusion, in this group of patients, the short-term clinical tolerance of the tricyclic antidepressant and fluoxetine association was very good, despite the pharmacokinetic interaction existing between these two types of drugs.