Same Witness, Different Arbitration: When Does Prior Exposure Put an Arbitrator’s Impartiality in Doubt?

An arbitrator had previously sat in a separate arbitration in which the same witness gave evidence and the tribunal ruled against parties connected with that witness. The arbitrator did not disclose that earlier involvement before presiding over a second arbitration in which the witness again gave evidence. Did that prior exposure, and the failure to disclose it, give rise to apparent bias?

In O v S [2026] HKCFI 4910, the Court of First Instance answered that question in the negative. The Court held that prior exposure to a witness, even if undisclosed and relatively recent, will not without more establish a real possibility of bias. Instead, the court will examine the full context, including the nature of the earlier proceedings, the tribunal’s reasoning, and whether the witness’s credibility was material to the outcome.

Background

The proceedings arose from an HKIAC-administered arbitration concerning an investment management agreement under which Party P managed US$117 million for Party D. Following termination of the agreement, Party D sought the return of the managed assets and damages for alleged breaches of contract and duties. A three-member tribunal issued a partial award substantially in Party D’s favour.

Party P subsequently discovered that the presiding arbitrator, Member R, had sat in an earlier arbitration involving entities connected to the corporate group founded and chaired by Witness Z, a witness called by Party P in the later arbitration. Witness Z had been the respondents’ sole witness in the earlier arbitration and had been cross-examined before the tribunal chaired by Member R. The earlier award contained findings adverse to the respondents (§§40–41).

Member R had not disclosed the earlier arbitration. Party P applied to the Court to set aside the partial award under section 81 of the Arbitration Ordinance (Cap. 609) and article 34 of the UNCITRAL Model Law.

Party P argued that Witness Z’s evidence was important to the arbitration. Party D had attacked his evidence as shifting, false and incredible, and had questioned the authenticity of relevant documents. Party P contended that Member R’s previous exposure to Witness Z, combined with his non-disclosure, gave rise to a real possibility of bias (§68).

The test for apparent bias

The Court (DHCJ Gary CC Lam) applied the established objective test: whether a fair-minded and informed observer, having considered all relevant facts, would conclude that there was a real possibility that the tribunal was biased (§47).

The notional observer is neither complacent nor unduly sensitive or suspicious. The observer considers both sides of the argument, appreciates the importance of context and reaches an objectively justified conclusion. There must also be a logical connection between the circumstances relied upon and the alleged departure from impartial decision-making (§§47–49).

The Court recognised that the private and confidential nature of arbitration heightens the importance of disclosure. A party may have no means of learning that an arbitrator has sat in an earlier arbitration involving overlapping participants or subject matter. Moreover, an arbitral award is subject to more limited review than a first-instance court judgment. These features place a “premium on frank disclosure” (§§50–51).

Under section 25 of the Arbitration Ordinance, an arbitrator must disclose circumstances likely to give rise to justifiable doubts about impartiality or independence. A failure to disclose may contribute to apparent bias, but it does not automatically establish a real possibility of bias or require the award to be set aside. It remains one circumstance to be considered in the overall assessment (§§52–55).

Prior criticism of a witness

The Court also held that mere fact that a judge or arbitrator has previously commented adversely on a witness, or found that witness unreliable, does not ordinarily establish apparent bias (§57).

The position may be different where the earlier decision-maker made exceptionally strong findings concerning the witness’s integrity, used unusually forceful language, or otherwise expressed views that could lead a reasonable observer to doubt the decision-maker’s ability to approach the witness’s evidence with an open mind. Personal friendship or animosity, close acquaintance with a central witness, or the expression of an unbalanced and firmly concluded view may also be relevant (§§58–62).

What did the tribunal actually decide?

The Court examined the reasoning of the partial award in detail. Among other things:

• Although Party D had vigorously attacked Witness Z’s credibility, the tribunal did not find that he was dishonest or unreliable, either expressly or by implication (§70(4)).
• In interpreting the investment management agreement, the tribunal relied primarily on the contractual wording, documentary evidence, commercial common sense and evidence from another witness. It did not reject Party P’s interpretation on the basis that it disbelieved Witness Z (§70(1)).
• The tribunal described it as “puzzling” that Witness Z, despite his apparent familiarity with the relevant bond, could not provide further details or supporting documents. The Court regarded this as a fair observation that any reasonable tribunal could have made. The absence of further documentation was itself sufficient to support the comment, without implying dishonesty (§§30, 70(4)).
• Certain findings depended on the documentary record and Party P’s failure to respond to repeated requests for information, rather than on any rejection of Witness Z’s evidence (§70(2)).

In gist, the Court concluded that a fair-minded and informed observer reading the award would therefore find neither an adverse determination concerning Witness Z’s honesty nor any indication that the tribunal relied upon information obtained in the earlier arbitration (§70(4)).

Non-disclosure did not establish a real possibility of bias

The Court acknowledged Party P’s difficulty as the earlier arbitration was confidential, Party P could not ascertain precisely what evidence Member R had heard or what views he had formed about Witness Z and the related entities. The relative proximity of the two arbitrations also meant that Member R’s earlier involvement could not be treated as remote history (§§68(3)-(4), 71(1)-(5), 72).

The Court also accepted that Member R’s answer to the challenge appeared argumentative. However, defensiveness did not necessarily indicate that he had something to conceal. It could equally reflect concern for his professional reputation and integrity (§§71(6)–(8), 73).

Taken cumulatively, these matters might create doubt. However, the test was not whether some doubt existed but rather whether there was a real possibility of bias.

Having regard particularly to the reasoning in the partial award and the absence of any material credibility finding concerning Witness Z, the Court held that the threshold was not met (§§71-73).

Apparent bias alone may not justify setting aside

Further, even if apparent bias had been established, the Court found that Witness Z’s credibility did not affect the result. The tribunal’s conclusions rested primarily on the contractual language, documentary evidence and inherent commercial probabilities. The outcome would therefore have been the same regardless of any view taken of Witness Z. This provided an independent ground for refusing to set aside the award (§75).

Key takeaways

The decision provides four practical lessons:

4. Prior exposure to the same witness is not enough. Something more is required, such as extreme earlier criticism, personal animosity, prejudgment or another logical basis for doubting the arbitrator’s ability to consider the evidence afresh.
5. Disclosure obligations are fact-sensitive. There is no bright-line rule requiring disclosure whenever an arbitrator has previously encountered the same witness. Whether disclosure is required depends on the nature of the prior involvement, the findings made, and whether the circumstances are capable of giving rise to justifiable doubts as to impartiality. However, the Court suggested that prior findings seriously impugning a witness’s integrity may require disclosure.
6. The award’s reasoning is critical. In deciding the issue of apparent bias, the court will examine all the circumstances, including whether the tribunal actually made and relied upon an adverse credibility finding, not merely whether the opposing party attacked the witness.
7. Prejudice must be addressed. Even where apparent bias is established, an award may stand if the alleged bias could not have affected the outcome.

The application was dismissed and Party P was ordered to pay costs on an indemnity basis. The decision reflects Hong Kong’s pro-arbitration approach while preserving a rigorous and context-sensitive safeguard against genuine arbitral partiality.

A timely discussion on arbitral appointments

The decision also resonates with a timely discussion hosted by Tanner De Witt the same day in August 2026: “Who gets appointed as an arbitrator and why?”

The panel brought together representatives of leading arbitral institutions to discuss the practical factors influencing appointments, institutional decision-making, and steps being taken to prevent arbitrator-bias situations. Against that backdrop, this judgment is a useful reminder that the appointment process involves not only who is selected, but also the disclosures and safeguards necessary to maintain confidence in an arbitrator’s independence and impartiality.

Read our post on the panel discussion here.