Unexpected Clinical Features of the Tarasoff Decision:
The Therapeutic Alliance and the "Duty to Warn"

Lawson R. Wulsin, M.D., Harold Bursztajn, M.D., and Thomas G. Gutheil, M.D.

The authors present a case report and discuss the clinical effects of the Tarasoff decision on the therapy of a potentially violent patient. They emphasize that the patient's ambivalence toward the intended victim can be used to foster the therapeutic alliance. The therapist's legal duty to the victim and therapeutic duty to the patient, they assert, can then be synergistically applied with an unexpected benefit: the patient's capacity to make choices is enhanced. (Am J Psychiatry 140:601-603, 1983)

Received June 21, 1982; revised Nov. 10, 1982; accepted Dec. 3, 1982. From the Massachusetts Mental Health Center. Address reprint requests to Dr. Wulsin, Department of Psychiatry, Massachusetts Mental Health Center, 72-76 Fenwood Rd., Boston, MA 02115. Supported in part by NIMH grant MH-16460. The authors thank Mses. Jean Jackson and Elyse Littaye for assistance in the preparation of this manuscript.

Since the California Supreme Court enunciated, in the 1975 Tarasoff decision [1], the principle of the therapist's special duty to protect third parties, critics of that decision have repeatedly predicted the countertherapeutic effects of this decision on the patient, society, and the therapeutic alliance. As Stone [2] noted,

The imposition of a duty to protect, which may take the form of a duty to warn threatened third parties, will imperil the therapeutic alliance and destroy the patient's expectation of confidentiality, thereby thwarting effective treatment and ultimately reducing public safety.

Gurevitz [3], in criticizing the decision, also noted,

The duty to warn will increase rather than decrease the likelihood of violence and will bring about greater use of involuntary care

Counterposed to these criticisms are the arguments of Wexler [4], who proposed that the Tarasoff decision be turned to therapeutic advantage in a family systems context. He wrote,

A Tarasoff-type obligation [may prompt) a paradigmatic (or at least a pragmatic) shift in the treatment of interpersonal violence from an intrapsychic model to a model more interactionist in perspective. . . . In terms of its overall impact, then, Tarasoff may help rather than hinder therapy.

Wexler's speculations are in principle clinically sound and even reassuring; the clinician, however, is understandably disturbed by the concept of therapeutic approaches being disruptively influenced from remote judicial benches. In any case this last prediction of the effect of the Tarasoff decision on therapy invites empirical demonstration. We present here a clinical report that illustrates the influence of the "duty to warn" upon the treatment of one patient, together with some recommendations and conclusions.


Mr. A, a 20-year-old single man, was admitted to the day hospital at the Massachusetts Mental Health Center 18 months after stabbing a stranger in the neck. He had attempted assaults on his mother and on numerous hospital staff members, demonstrating a range of behaviors consistent with the DSM-III diagnoses of alcoholism and antisocial personality disorder. He often heard the voice of his deceased father saying, "Kill, kill!" and "Die, die!"; specifically, the voice commanded him to kill his mother. He refused to allow his family to be contacted or involved in his treatment, but he asked for help with the voices. He had no other sign of psychosis, and antipsychotic medication brought him little detectable relief. When Mr. A's hallucinations took the form of commands to kill his mother, we as part of the treatment staff became concerned about a possible duty to third parties. (Although Massachusetts has had no specific case "on point" for this issue, clinicians generally act as though the reasoning employed in Tarasoff applied here [5]. In this regard, we propose that our thesis in this article serve as a model for clinicians' behavior in ambiguous situations.) The treatment staff concluded that, in an "open ward" setting, a duty to warn did exist; transfer to a closed unit was rejected as too regressive. In keeping with the principle of maintaining the therapeutic alliance whenever possible [6], especially in legal matters [7], we elected to involve the patient maximally in the process. To this end, we proposed a draft of a letter that would inform Mr. A's mother of the danger to her and that would also serve to document our response to her son's threats. In keeping with an alliance-seeking approach, Mr A's therapist went over the letter and the attendant rationale with him. The letter stated that the patient "feared he might harm [his mother)." Mr. A agreed with the content of the letter and insisted on talking to his mother before we mailed the letter, fearing the letter would cause his mother to wish never to speak to him again. His mother first responded to the letter by saying that he should be "locked up with the key thrown away. "During the ensuing conversation, however, she stated openly, "I love you" ; Mr. A responded, "I love you, too," and both began to cry.

Thereafter, Mr. A abided by a temporary agreement with the therapist not to see his mother outside the treatment setting; but he continued telephoning her and the family every day. Although his mother volunteered information to us by telephone, she otherwise refused to participate actively in her son's treatment. No civil commitment or further intervention was necessary for Mr. A.


This vignette depicts a clinically based approach to the duty to protect third parties as enunciated in the Tarasoff decision and demonstrates that Wexler's position [4] can have clinical validity. The patient and the mother had interchanged the roles of potential assailant and victim many times in the patient's childhood. At the time of the threats these roles still appeared to be potentially interchangeable: each wanted to do away with the other. This interchangeability of roles reflected the intense ambivalence expressed by each for the other.

Given that trust is the sine qua non of the therapist-patient relationship and that confidentiality is the mechanism for protecting that trust, only rarely can the therapist breach the patient's confidence without losing the patient's trust (and the patient). This point has been extensively reviewed by Meisel and Roth [8], who underscore the importance of reviewing the breach with the patient.

In our case, several factors promoted the therapeutic alliance in spite of the breach of confidence. First, the informing process we employed-open, nondeceptive, actively including the patient-closely resembles the process used to obtain informed consent for a procedure Here, however, consent was obtained even when the issue was the patient's own dangerousness to another. The alliance, the optimal arena in which informed consent takes place, is strengthened by such a process; in addition, the attempt to obtain informed consent often serves as a test of the durability of the alliance and of the patient's ego capacities-both important factors in the assessment of dangerousness.

Second, our review of the letter with the patient capitalized on exploring the affective ambivalence the patient had toward his victim, as he himself demonstrated by actively participating in the informing process and choosing to be the bearer of this information to his mother. Finally, the breach of confidence demonstrated the clinicians' intent to protect the victim (and the patient) from serious harm as a result of the patient's violent impulses; the breach of confidentiality thus functioned like an external control.

The clinical context that gives rise to the issue of a Tarasoff duty contains an inherent paradox; the patient seems to act to thwart his own wishes. That is, the patient who intends harm informs the therapist, who has-in theory at least-some power to prevent that harm. The paradox of informing the therapist reflects the patient's fear of his own aggressive wishes and his ambivalence toward actually harming the victim. Continuing the process of informing the victim forces a graphic labeling of affects and moves them, as it were, back into the interpersonal context where they developed; more importantly, the patient talks to the intended victim instead of acting. If the warning is performed with explicit recognition of the patient's ambivalence about the intended harm, it allows the therapist to ally with the healthy part of the patient's ego that fears the assault.

As Appelbaum [5] has noted, one alternative for the clinician to the Tarasoff approach has traditionally been civil commitment for dangerousness, which has offered the advantages of maintaining confidentiality. In keeping with the spirit of Appelbaum's proposed options, our case vignette reveals empirically several therapeutic advantages of approaching the duty to warn in clinical terms, even with someone already hospitalized in an open setting. First, the patient's relationship with the victim is brought to center stage in the therapy; the warning process, as Wexler [4] had predicted, serves to pull the potential victim actively into the therapy (though only marginally and briefly in our case). Second, the patient can identify with the therapist's deliberate (verbal) approach to negotiating with the intended victim. In support of this view, our patient has, in fact, been able to interact verbally in other aggression-producing situations. Finally, the approach described may permit use of a "less restrictive therapeutic setting," where the alliance functions in place of an elaborate panoply of constraints and warnings to promote the patient's ability to work therapeutically with people, including particularly the intended victim.

The objection might be raised here that this "deliberate" approach poses undue risk for the potential victim. Clearly, the individual case circumstances must be considered: what is the level and immediacy of the risk? How much control over that risk can the clinician exercise (e.g., is the patient already in a hospital or not) ? The short- and long-term advantages of such a "cool" approach to decision making [9] seem to us to justify its use even in a crisis, as we have described, to the benefit of both patient and victim.


  1. Tarasoff v Regents of the University of California, 131 Cal Reptr 14 (Cal 1976)
  2. Stone AA: The Tarasoff decisions: suing psychotherapists to safeguard society. Harvard Law Review 90:358-378, 1976
  3. Gurevitz H: Tarasoff: protective privilege versus public peril. Am J Psychiatry 134:289-292, 1977
  4. Wexler D: Patients, therapists and third parties: the victimological virtues of Tarasoff. Int J Law Psychiatry 2:1-28, 1979
  5. Appelbaum PS: Tarasoff: an update on the duty to warn. Hosp Community Psychiatry 32:14-15, 1981
  6. Gutheil TG, Havens LL: The therapeutic alliance: contemporary meanings and confusions. Int R Psycho-Anal 6:467-481, 1979
  7. Gutheil TG, Appelbaum PS: Clinical Handbook of Psychiatry and the Law. New York, McGraw-Hill, 1982
  8. Roth LH, Meisel A: Dangerousness, confidentiality and the duty to warn. Am J Psychiatry 134:508-Si 1, 1977
  9. Janis IL, Munn L: Decision Making. New York, Free Press, 1977