Non-Defendant Bad Character (CJA 2003 s.100): Excluding Hearsay Allegations and Policing “State of Mind” Evidence to Prevent Backdoor Propensity Proof

1) Introduction

R v Horswill ([2026] EWCA Crim 181) concerned an appeal against a murder conviction following a fatal strangulation in the garden of the deceased, Richard Langley. The appellant’s defence was self-defence: he claimed he went to the deceased’s home after an earlier pub altercation and was then attacked with a weapon, responding with force to protect himself.

The appeal focused on non-defendant bad character evidence (the deceased’s alleged past violence, convictions, and community “reputation”), which the defence sought to deploy for two purposes:

  1. Propensity: to show the deceased was more likely to be the aggressor (including allegedly using a weapon).
  2. State of mind: to explain why the appellant turned back to the deceased’s home and (to some extent) to bolster the appellant’s credibility as to what he feared and why.

The trial judge admitted two incidents by agreement (October 2019 headbutting; September 2021 escalating confrontation including going to a person’s house and assault) but excluded a wider set of disputed materials (including old convictions and largely hearsay allegations). The Court of Appeal (Criminal Division) dismissed the appeal, holding the trial judge’s approach was lawful and, in any event, the conviction was safe.

2) Summary of the Judgment

  • The disputed material was not admissible as important explanatory evidence under Criminal Justice Act 2003, s.100(1)(a).
  • Considering admissibility under s.100(1)(b), the trial judge was entitled to reject the disputed evidence as lacking substantial probative value, particularly where it was hearsay, ill-defined, untested, or risked satellite litigation.
  • The Court emphasised the need to identify clearly the issue to which non-defendant bad character is said to go (here: propensity vs state of mind) and to evaluate probative value in the context of what was already before the jury.
  • On “state of mind”, the appellant had not clearly articulated which specific incidents he knew about and how they affected his decision-making; the judge was entitled to prevent the defence using “state of mind” as a route to circumvent her earlier bad character ruling.
  • Even if error were shown, the Court held the conviction was plainly safe (Criminal Appeal Act 1968, s.2), given CCTV threats, forensic evidence, and other trial features.

3) Analysis

3.1 Precedents Cited

R v Braithwaite [2010] EWCA Crim 1082

R v Braithwaite [2010] EWCA Crim 1082 was central. The Court of Appeal in Horswill drew and applied multiple points of guidance taken from Braithwaite, notably:

  • “Substantial probative value” under s.100 is a demanding test and is not the same as the more familiar defendant bad character gateway analysis.
  • Where s.100 conditions are met, there is no residual discretion to refuse admission; conversely, where they are not met, evidence is inadmissible.
  • The judge must identify the issue the evidence addresses and decide whether (i) the issue is of substantial importance in the case and (ii) the evidence has substantial probative value on it.
  • Hearsay allegation evidence will “rarely” be of substantial probative value, because of the jury’s difficulty assessing it, particularly where complainants do not support allegations or police did not pursue them.

In Horswill, Braithwaite underpinned the rejection of disputed materials such as crime reports, unsupported allegations, and hearsay accounts (e.g., those not formally complained of, not prosecuted, lacking witnesses, context, dates, or detail). The Court also endorsed the trial judge’s “satellite litigation” concern: admitting contested, under-particularised incidents would have shifted the trial into mini-trials about old and unclear events.

R v BVY [2024] EWCA Crim 1355

R v BVY [2024] EWCA Crim 1355 supplied the appellate restraint standard. Whether evidence has substantial probative value is a judgment call for the trial judge. The Court in Horswill reiterated it will not interfere unless:

  • there was a misdirection on legal principle, or
  • the evaluation was plainly wrong (one no reasonable judge could reach).

Applying BVY, the Court treated the trial judge’s exclusion of aged convictions, ill-defined allegations, and hearsay as firmly within the permissible evaluative range—especially because the jury already had significant admissible evidence of the deceased’s violent conduct.

R v Hussain [2008] EWCA Crim 1117

R v Hussain [2008] EWCA Crim 1117 was deployed to clarify the difficult boundary between:

  • evidence of a defendant’s belief (relevant to state of mind), and
  • evidence offered to prove another person’s misconduct (bad character within the statutory regime).

Hussain recognised that a defendant may be entitled to describe what he believed about another’s past and—where there is some factual foundation—adduce evidence of that foundation to support (i) truthfulness of the asserted belief, and (ii) reasonableness of the defendant’s reaction (there, in a duress context).

In Horswill, however, the Court stressed two counterweights:

  1. Trial judges must be astute to prevent “state of mind” from becoming a device to smuggle in inadmissible bad character (particularly where a s.100 ruling has already excluded the material).
  2. The defence must articulate with clarity what the defendant knew, what incidents he had heard about, and how that knowledge bore on the state of mind issue; here, the application and case statements were treated as a “moving target” and did not properly link disputed incidents to the appellant’s knowledge and decision to go to the deceased’s home.

Thus, while Hussain demonstrates that “belief” evidence may sometimes be admissible on a non-bad-character footing, Horswill shows courts will require disciplined pleading and careful relevance analysis, and will not permit the approach to be used to circumvent s.100.

3.2 Legal Reasoning

(a) The statutory framework: CJA 2003, ss.98 and 100

The Court set out the statutory definitions and gateways:

  • Section 98 defines “bad character” as evidence of, or of a disposition towards, misconduct, excluding evidence directly to do with the alleged facts of the offence or misconduct in the investigation/prosecution.
  • Section 100 strictly limits admissibility of non-defendant bad character to:
    • s.100(1)(a) important explanatory evidence;
    • s.100(1)(b) substantial probative value on an important issue; or
    • s.100(1)(c) agreement of all parties.

(b) Important explanatory evidence rejected (s.100(1)(a))

The Court held the disputed evidence did not meet either limb of the “important explanatory evidence” test: the jury could understand the case without it, and its value for understanding the case as a whole was not substantial. Accordingly, admissibility—if any—had to be justified under s.100(1)(b).

(c) Separating the “propensity issue” from the “state of mind basis”

A key analytical move was the Court’s insistence (from Braithwaite) on identifying the issue to which evidence goes. The Court treated propensity and state of mind as distinct and assessed each separately. This mattered because evidence potentially strong on propensity might be weak on state of mind (if unknown to the defendant), and evidence potentially relevant to state of mind might not meet s.100 (if in truth offered to prove misconduct).

(d) Propensity: probative value assessed against what was already admitted

The Court upheld the trial judge’s refusal of disputed evidence largely because the jury already had significant, concrete evidence bearing on propensity:

  • the accepted October 2019 headbutting (deceased as aggressor),
  • the accepted September 2021 escalation (including returning later to a home and assaults, including on a partner), and
  • evidence about the confrontation with the appellant’s partner on the night in question.

Against that evidential backdrop, the disputed evidence was (i) old (e.g., 2005; 2014), (ii) different in nature (the 2005 sexual offence), (iii) thin in detail, or (iv) untested hearsay allegations with non-prosecution/no complaint features—precisely the category Braithwaite warned would rarely have substantial probative value. The Court also approved the judge’s concern about presenting the 2005 conviction to the jury on a misleadingly “violence-only” description.

The Court also firmly rejected the defence submission that alleged drug dealing itself could support a propensity for violence as “completely hopeless”, illustrating that s.100 requires probative value grounded in evidence, not stereotype.

(e) State of mind: the need for clear linkage and the anti-circumvention principle

On appeal, the defence sought to characterise excluded materials as relevant to why the appellant turned back to the deceased’s home and to bolster credibility. The Court’s response was twofold:

  1. Under-specification problem: the appellant did not clearly identify what he had heard, what incidents he knew of, and how these affected his decision-making. The Court considered the defence approach “a moving target”, leaving the judge without a coherent basis to admit disputed incidents on a state-of-mind foundation.
  2. Policing circumvention: the trial judge was entitled to intervene when questioning threatened to open the door to the very bad character evidence she had already excluded under s.100. The Court endorsed the judge’s proportionate solution: the appellant was allowed to give limited evidence that he acted because of the deceased’s “reputation for not letting things go,” and that this concern related to his partner and daughter, but not to expand into wider inadmissible allegation evidence.

Importantly, the Court noted that the one admitted incident that might objectively support a “won’t let things go” reputation was the September 2021 escalation, but there was no evidence that the appellant actually knew of it or that it influenced him. This illustrates the Court’s core point: “state of mind” relevance depends on the defendant’s knowledge and the evidential foundation for that knowledge being clearly explained and appropriately admissible.

(f) Summing-up complaint rejected; safety of conviction affirmed

A late summing-up complaint (that the judge should have directed on state-of-mind relevance of the two admitted incidents) was rejected because it was not raised at trial, was not grounded in evidence of the appellant’s knowledge, and was not realistically arguable.

The Court then addressed safety under Criminal Appeal Act 1968, s.2. Even if the evidential rulings were wrong, the conviction was safe given: CCTV threats to kill; pathology undermining the appellant’s account of strangulation; the deceased’s defensive injuries; the appellant fleeing; and the adequacy of the self-defence direction.

3.3 Impact

  • Structured s.100 analysis is mandatory: courts should insist that parties identify with precision whether non-defendant bad character is advanced for propensity, state of mind, or another issue, and then apply s.100(1)(b) to each issue distinctly.
  • Hearsay and “allegation bundles” remain hard to admit: Horswill reinforces Braithwaite’s scepticism, especially where there is no prosecution, no contemporaneous complaint, no clear source, or the allegation would trigger satellite litigation.
  • State of mind is not a backdoor gateway: trial judges are entitled to curb questioning framed as “reputation/belief” where, in substance, it seeks to introduce excluded bad character; defendants must clearly plead what was known and why it mattered.
  • Appellate deference on probative value: BVY is applied to protect reasonable trial-level evaluative decisions, making appeals difficult absent clear misdirection or irrationality.

4) Complex Concepts Simplified

Bad character (CJA 2003 s.98)
Evidence that someone has behaved badly in the past (misconduct), used to support an inference about their behaviour or credibility.
Non-defendant bad character (CJA 2003 s.100)
Bad character evidence about someone other than the accused (e.g., the deceased). It is admissible only through limited statutory “gateways”.
Important explanatory evidence (s.100(1)(a))
Evidence without which the jury would find it difficult or impossible to understand other evidence, and which materially helps to understand the case as a whole.
Substantial probative value (s.100(1)(b))
Evidence that significantly helps prove or disprove an important issue in the case. It is a high threshold, especially for untested allegations.
Propensity
The idea that a person has a tendency to behave in a particular way (e.g., to initiate confrontations), making it more likely they acted that way on the occasion in question.
State of mind
What the defendant believed or feared at the time, relevant to issues such as self-defence (e.g., why the defendant acted as they did). Relevance often depends on what the defendant actually knew.
Hearsay
Second-hand accounts offered for their truth (e.g., “someone told me he did X”), generally weaker because it cannot be tested in cross-examination.
Satellite litigation
Side disputes that risk distracting the jury with mini-trials about collateral incidents, undermining fairness and focus.

5) Conclusion

R v Horswill confirms a disciplined approach to non-defendant bad character under CJA 2003 s.100: allegation-heavy, hearsay-laden or ill-particularised material will rarely meet the substantial probative value test, especially when the jury already has concrete admissible evidence on the same issue. The decision also provides practical guidance on “state of mind” arguments: defendants must clearly identify what they knew and why it mattered, and courts may properly restrict questioning that would otherwise circumvent an adverse s.100 ruling. Finally, the judgment illustrates that even where evidential disputes arise, appellate relief depends on whether any error renders the conviction unsafe—a threshold not met on these facts.