Ward v Donnellan: when both sides lie, "no order for costs" is not a free pass

There is a certain judicial instinct, when both sides in a trial have been caught lying, to throw up one's hands and order that everyone bears their own costs. It feels fair. It punishes both liars equally. Ward & Ors v Donnellan & Ors [2026] EWCA Civ 729 is a useful corrective to that instinct, and a reminder that "no order for costs" is a conclusion that still has to be reasoned, not a shortcut around the general rule.

The underlying trial, before Louise Hutton KC in the Business and Property Courts, ran for fifteen days and produced a judgement of 473 paragraphs. At its heart was a claim by Anthony Donnellan that he and Alan Ward had formed a partnership to develop a building in Battersea, with related claims about who held various flats on trust. Donnellan lost entirely. The judge found no partnership and no joint venture ever existed, and that Donnellan's account of key meetings supporting his case was simply made up. Ebonair's counterclaim, that two individuals holding flats in the building did so on trust for it, succeeded as a direct consequence.

Ward did not emerge unscathed either. The judge found he had relied on sham documents, a loan agreement and trust deed she concluded were fabricated years after their purported dates, to hide assets from his trustee in bankruptcy. He also lost on the question of who beneficially owned Ebonair, the company at the centre of the dispute.

At the costs hearing, the judge made no order for costs on both the partnership claim and the related Part 20 claim, reasoning that Ward's dishonesty meant he should not benefit from costs incurred advancing a "dishonest case", and that this roughly cancelled out what Donnellan would otherwise owe as the unsuccessful party.

Lord Justice Lewison, giving the leading judgement, found this approach wrong in principle, and the reasoning is worth dwelling on. The starting point under CPR 44.2 remains that the loser pays. Dishonesty by the winning party justifies stripping out the costs of advancing that dishonest element, and compensating the other side for the cost of exposing it, but it does not automatically wipe out everything else, including the costs the winner reasonably incurred defending a claim that ultimately failed in its entirety.

The court drew a sharp distinction from Intrigue Shipping v Nikitin, where a "no order" outcome was upheld because the claimant had failed on its primary case, which had consumed most of the trial's cost and effort. Here, by contrast, Ward succeeded on everything that mattered. His dishonesty over Ebonair's ownership was, on analysis, largely irrelevant to whether a partnership existed at all. The judge had effectively allowed a side issue to swallow the main event.

There is a structural lesson here too. Where multiple actions are tried together, as happened with the partnership claim, the Part 20 claim and the possession claim, dishonesty findings relevant to one claim cannot simply be imported wholesale into the costs analysis of another. Ward's dishonesty about Ebonair had nothing to do with whether Keane and Howard held their leases on trust for Ebonair under the Part 20 claim, yet the judge treated the two as inseparable. The Court of Appeal also flagged a basic inconsistency: the Part 20 claim and the possession claim stood or fell together, yet attracted contradictory costs treatment.

The result was a costs order of its own: Donnellan to pay 50% of Ward's costs of the partnership claim, and Keane and Howard to pay Ebonair's costs of the Part 20 claim. For anyone drafting costs submissions after a trial featuring credibility findings against multiple parties, the message is to separate the issues, not the personalities, and to keep the loser pays starting point firmly in view even when nobody comes out of the judgement looking good.