There is a consensus in the psychotherapy research field to consider the therapeutic alliance, broadly defined as the mutual collaboration between the therapist and the patient, as a robust and consistent predictor of therapy outcome. There is little agreement, however, on the best way to operationalise and measure it. Several instruments are available, each of them displaying some unique features so that investigators have problems to choose among them on the basis of considerations other than ease of administration, continuity with previous research or availability. One of the early self-report measures and widely used instruments was the Helping Alliance questionnaire (HAq-I) (Alexander and Luborsky, 1986). In recent years, we have become aware that it was limited by the presence of items that were explicitly assessing early symptomatic improvement and by the fact that all the items were worded positively. A revised version (HAq-II) was developed (Luborsky et al., 1996), guided by two main goals: (1) to reduce the inclination of the scale toward measuring early symptomatic improvement and thus confusing these two dimensions, and (2) to better incorporate the various aspects of the alliance related to the collaborative effort of patient and therapist. The new instrument includes 5 from the 11 items of the HAq-I and 14 new items - 5 of them worded negatively. The aim of the study is twofold: to validate the French version of the new HAq-II and to investigate empirically to what extent the HAq-II has improved over the HAq-I. The sample included 60 self-referred outpatients assigned to a Brief Psychodynamic Investigation (BPI), a manual-based investigation procedure in four sessions guided by psychodynamic principles. We looked at the correlation between the two HAq (I and II) and the Working Alliance Inventory (WAI), patient pretreatment characteristics (SCL-90, HDRS, HAMA, IIP) and outcome (SCL-90, SAS and patient satisfaction). Results showed that the French version of the HAq-II has good psychometric properties. Estimates of internal consistency and test-retest reliability were fairly similar to the original English version. Indication of its validity included high correlation with other alliance measures and independence from patient pretreatment characteristics. Surprisingly, HAq-II score predicted patient's satisfaction with the treatment but not symptomatic improvement. Taken together, these first results are promising and indicate that the translated version of the HAq-II is a valid instrument for measuring the helping alliance. Concerning the comparison between the two versions of the HAq, HAq-II has proved to be an improvement compared to the original HAq-I scale: it better relates to the alliance construct and it is less influenced by the symptoms of the patient. Considering also its better construct validity (Luborsky et al., 1996), we definitely recommend the use of the revised HAq-II instead of the initial version of the scale. (PsycINFO Database Record (c) 2006 APA, all rights reserved)