In there general practice, physicians use magnesium in the treatment of somatic manifestations of anxiety. Magnesium is involved as cofactor of approximately 300 enzymatic reactions, and more over as it reduces the stimulating effect of L-glutamate on N.M.D.A. receptors (Mayer, 1994). This fact suggests that it may have pharmacological effects. Some authors have suggested that it could have an effect comparable to lithium on [ldquo ]rapid bipolar affective disorders[rdquo ] (Chouinard et al, 1990). In another works, it is suggested that magnesium could prevent the premenstrual syndrome (Rosenstein et al, 1994).The anxiolytic effect of PCR 7060, a combination of magnesium lactate and pyridoxine, has been assessed in two open studies vs. lorazepam (Bourgeois et al, 1987; Scharbach et al, 1988). We therefore decided to compare PCR 7060 to Buspirone in a double-blind controlled study on patients with Generalised Anxiety Disorder, as defined by DSM III-R.Method: Following a 7 day placebo wash-out period, 103 patients with medium intensity G.A.D. (Hamilton Anxiety Scale score between 15 and 30) were randomised into two treatment groups: PCR 7060 (200 mg Mg + + and 20 mg pyridoxine) or Buspirone (20 mg). Changes in patients condition were assessed according to the Hamilton Anxiety Scale (HARS). PCR tolerance was evaluated according to CHESS scale, on D7, D10, D21 and D42 i.e. throughout the 6 weeks treatment.Results: 99 out of 103 patients were included on D0. Their HARS score was above 15.Amongst the 15 patients int the PCR group, 12 were prematurely withdrawn from the study (3 due to ineffective treatment and 1 due to intolerance), and 12 amongst the 48 patients in the Buspirone group (3 due to ineffective treatment, 1 due to intolerance, and 2 for both reasons).The decrease in the average HARS score in the PCR group (21.8 [plusmn] 4.5 on D0; 14.9 [plusmn] 5.9 on D21 and 11.7 [plusmn] 5.2 on D42) was not statistically different (p = 0.52) to the one seen in the Buspirone group (21.3 [plusmn] 3.9 on D0; 17.3 [plusmn] 6.6 on D21 and 13.8 [plusmn] 5.6 on D42). Similarly, the changes in V.A.S. observed in both groups were not statistically different (p = 0.95).However, PCR had a greater effect on the symptoms of HARS's somatic factor than Buspirone: 10 [plusmn] 2.6 vs. 10.4 [plusmn] 3.1 on D0; 6.9 [plusmn] 3.1 vs. 8.6 [plusmn] 4 on D21 (p = 0.03) and 5.5 [plusmn] 3.1 vs. 7 [plusmn] 3.5 on D42 (p = 0.06).Moreover, the frequency of the adverse effects screened by CHESS was significantly lower (p = 0.04) in the PCR group than in the Buspirone group (31%).Discussion -- Conclusion: This study showed that the anxiolytic effect of PCR 7060 is not statistically different to that of Buspirone in the treatment of patients with G.A.D. However, it demonstrated that PCR tolerance was greater and its action on the somatic expression of anxiety was superior to that of Buspirone.In this present study: no statistical difference between PCR 7060 and Buspirone in the treatment of patients with G.A.D. disorder.Fewer side effects have been reported in PCR 7060 group than in Buspirone group.