Working Alliance Inventory
Working Alliance Inventory (WAI) · Also known as: WAI, WAI-36, WAI-SF, WAI-SR
The Working Alliance Inventory (WAI) is a validated, empirically supported measure of the therapeutic alliance—the collaborative relationship between therapist and client. Developed by Horvath and Greenberg in 1989, the WAI operationalizes Bordin's tripartite model of alliance: agreement on goals, agreement on tasks, and emotional bond. It is one of the most widely used alliance measures in psychotherapy research and is a strong predictor of psychotherapy outcome across diverse theoretical orientations and client populations.
Read the full method
Sign in with a free account to read this section.
Method map
The neighbourhood of related methods — select a node to explore.
When to use it
The WAI is recommended for: (1) individual psychotherapy outcome monitoring in clinical practice and research; (2) early treatment engagement assessment (after session 1–3); (3) identifying alliance ruptures for timely intervention; (4) comparing alliance across therapeutic modalities (CBT, psychodynamic, humanistic); (5) training therapists to attend to the relationship; (6) psychotherapy research examining alliance as a mediator or predictor of outcome. It is not specific to a single diagnosis or age group—it has been validated across child, adolescent, and adult populations, though child and adolescent versions require adapted language.
Strengths & limitations
- Strongest empirical support for alliance measurement; validated across 50+ studies and multiple therapy types.
- Predictive validity: early alliance predicts treatment engagement, retention, and outcome in psychotherapy.
- Bidirectional (therapist and client perspectives); therapist–client agreement is itself clinically informative.
- Short form (WAI-SF, 12 items) is efficient for repeated measurement and routine outcome monitoring.
- Transdiagnostic; not tied to specific presenting problems, allowing comparison across heterogeneous caseloads.
- Tripartite structure (Goal, Task, Bond) operationalizes theory-informed dimensions of alliance.
- Therapist and client ratings often diverge; therapists typically rate alliance higher than clients, which can mask emerging ruptures.
- The 7-point scale has low endpoint discrimination; few respondents select 1 or 7, concentrating data in the middle range.
- Cross-cultural validation is limited; most development and validation work is North American/European.
- Administration requires respondent literacy and willingness to complete Likert scales; less suitable for very low-literacy populations.
- Alliance is state-dependent; scores reflect the present moment and may not generalize to other sessions or therapeutic phases.
- No validated shorter form (e.g., 4-item screener) widely adopted; the 12-item SF is minimal standard.
Frequently asked
What is a 'good' WAI score?
On the 12-item short form (range 12–84), scores above 60 are generally considered indicative of solid alliance. In outcome research, average scores above the median (~70–72) are associated with better outcomes. However, alliance is dynamic; a low session-1 score does not predict poor outcome if alliance strengthens by session 3. Use trend over single scores.
Why do therapist and client WAI ratings often disagree?
Therapists tend to rate alliance higher than clients; this discrepancy is normal and clinically meaningful. Large gaps (>15 points) suggest unspoken tension, goal misalignment, or client reluctance. Therapist over-optimism about alliance does not protect against dropout—use the lower rating (client's perspective) as the more predictive signal.
How often should I administer the WAI?
Standard practice is: (1) after session 1 (baseline, identifies early alliance challenges); (2) after session 3–5 (early alliance predictor of outcome); (3) then monthly or session-by-session in routine outcome monitoring. In research, every session or every third session is typical. More frequent measurement increases sensitivity to rupture but may burden respondents.
Can the WAI be used for couples therapy, family therapy, or group therapy?
The WAI was developed for individual dyadic therapy. For couples therapy, administer separately to each partner regarding their relationship with the therapist. For family/group therapy, the alliance structure becomes complex (alliance with each family member, with the group), and alternative measures (e.g., group climate scales) may be more appropriate. Individual adaptation studies exist but are less established.
Is a high WAI score sufficient for good outcomes?
No. Strong alliance is necessary but not sufficient. Outcomes also depend on treatment content, therapist skill, client motivation, and problem fit. A strong alliance with ineffective interventions will not produce good outcomes. Alliance is a foundation; technical competence and appropriate case conceptualization are also essential.
How do I address a low WAI score therapeutically?
If client WAI is low or dropping, use alliance repair: (1) acknowledge the client's experience directly ('It sounds like this doesn't feel helpful'); (2) explore discrepancy (clarify goals: 'What brought you in, and is that still what you want?'); (3) renegotiate tasks ('Would it be useful to shift focus or approach?'). These conversations often strengthen alliance and prevent dropout.
Sources
- Horvath, A. O., & Greenberg, L. S. (1989). Development and validation of the Working Alliance Inventory. Journal of Counseling Psychology, 36(2), 223–233. DOI: 10.1037/0022-0167.36.2.223 ↗
How to cite this page
ScholarGate. (2026, June 3). Working Alliance Inventory (WAI). ScholarGate. https://scholargate.app/en/psychotherapy-research/working-alliance-inventory
Which method?
Set this method beside its closest kin and read them side by side — the library lays the books on the table; the choice is yours.
- Collaborative Study Psychotherapy Rating ScalePsychotherapy Research↔ compare
- Outcome Rating ScalePsychotherapy Research↔ compare
- Session Rating ScalePsychotherapy Research↔ compare
- Therapeutic Alliance ScalePsychotherapy Research↔ compare