The Morning It Becomes Real
During a break in court proceedings a client and his senior counsel stand in the corridor, waiting for the matter to be called. Passing the time, the client asks the silk how many years of experience he has in court. The silk smiles faintly. “In court? Five years. In the corridor, twenty-five.”
The joke survives because it is true. In the Gauteng Division of the High Court about eighty five per cent of civil trials settle on the morning of the hearing, most of them in that corridor. The Division has now made mediation compulsory. A directive from the Judge President requires the parties in civil matters to mediate before trial, and from January 2027 no trial date will be given without a mediator’s report. Whether the directive will do what it was built to do is a separate question, and the answer decides something practical, which is how much of a client’s money should be spent on a step the client has no choice about.
The Morning It Becomes Real
Why mandatory mediation will not change when matters settle in Gauteng
Anyone who has run civil trials in this Division knows the shape of the first day of trial. The client arrives early, in a suit he does not wear often, and he is quieter than he has been in four years of correspondence. His witnesses wait in the corridor and one of them looks less certain than he did in consultation. Counsel is in her robes. Somewhere along the same corridor the other side stands in a group of its own, doing the same sums, what the next five days will cost and what number would make the whole thing go away.
At ten o’clock one or both counsel ask for the matter to stand down until after tea so that the parties can talk. By half past ten, more often than not, the matter is settled, and when it is called again a surprisingly short settlement agreement is handed up and made an order of court, by agreement between the parties. The silk in the joke was not only being witty. He was telling the client where matters like his are actually decided.
When the directive was under consultation the eighty five per cent figure was quoted in almost every commentary, and always to make the same point. If these matters could settle on the morning of trial, they could have settled years earlier, and compulsory mediation early in the case would capture the same settlements at a fraction of the cost. With trial dates that had run out to 2031, the argument was hard to resist.
In my experience the argument is wrong. Those matters settle on the morning of trial because of that morning. The pressure that produces the settlement is created by the occasion itself, and a meeting held in a boardroom eighteen months earlier, with nothing yet at stake, cannot reproduce it.
Why do matters settle so late?
Settlement is a function of pressure and information, and on the morning of trial both arrive at once. The cost of the next five days stops being an estimate in a letter and becomes a certainty. The witnesses are no longer names on a list but people about to be cross-examined, and counsel has just watched one of them go pale. The client also now sees the room where the duel will take place, and courtrooms are designed to impress and intimidate. Before he came to court, the client was told for four years that he has a strong case, and nothing has yet tested that belief. Now a judge is about to test it, in public, and the client discovers that he would prefer not to find out.
The advice changes as well, because the question has changed. An opinion written eighteen months before trial deals with prospects, and prospects come in ranges and rest on assumptions. On the morning of trial the client asks a different question. He asks whether he should walk into that room, and the answer comes back in a sentence, shorter and harder than anything in the opinion.
There is a second reason for these settlements. A good deal of litigation is cathartic. The parties are angry with each other, and for years the pleadings and the correspondence give the anger somewhere to go, so that what is being pursued is closer to vengeance than to justice. By the morning of trial the anger has done its work and pragmatism takes its place, helped along by the sober advice of counsel, who has prepared the case in the weeks before and found the cracks in the armour of what everyone had been calling an unloseable case.
None of that is a failure of facilitation. Offers had been made and refused, without prejudice correspondence had passed for years, and settlement had been discussed at a pre-trial conference and gone nowhere. What was missing was never the opportunity. It was the pressure, and nothing supplies the pressure except proximity to a decision.
A litigant made the point rather well. In the case of Brondani[1], decided in Johannesburg in November 2025, a defendant resisted an application to compel his co-operation in appointing a mediator. He said that the parties had already tried to settle and failed, that the acrimony between them made the exercise pointless and that the costs would be wasted. Adams J ordered him to mediate, holding that a litigant’s belief, however genuinely held, cannot decide whether a matter goes to mediation, and that acrimony and a subjective view that the process will waste time are not exceptional circumstances.
The court was right and so was the litigant, but about different things. A party cannot be the judge of whether he must attend mediation, because if he could, the directive would mean nothing. But he was probably right that attending would not produce a settlement, and the judgment does not suggest otherwise. It decides that he must go, not that it will work.
What is the directive assuming?
The directive assumes that what these matters have been missing is a structured opportunity to negotiate, run by a neutral person trained for the work. In many fields that is a fair assumption. It does not describe commercial litigation, where the parties are represented throughout, talk constantly and have usually been negotiating for years. What is missing is not a forum. It is information, mandate and pressure.
The timing makes it worse. The mediation is initiated by a notice under rule 41A of the Uniform Rules of Court, expanded by the directive to make each side state its position in writing, and it runs on deadlines measured in court days. All of this happens near the beginning of a matter. Discovery has not taken place. Expert reports usually do not yet exist. In a construction matter or a professional negligence claim, neither side yet knows within a wide margin what the case is worth.
There is a class of matter in which the assumption fails altogether. Where the defendant is an organ of state, a large insurer or anyone whose settlement mandate comes from a committee that does not sit in the room, the timing is beside the point. The person who attends cannot settle in January for the same reason that he will not be able to settle in November. The decision is not his to take. Compelling those parties into a room earlier produces an earlier failure, not an earlier settlement.
So the directive requires the parties to negotiate at the point in the life of a matter when they know least about it. That is not merely an unhelpful moment. It is close to the worst one, because a defendant who does not know his exposure cannot obtain a mandate for it, and a plaintiff who cannot value her claim has nothing to compromise. The parties are being asked to do at eighteen months what they will be perfectly capable of doing on the morning of the trial, when everything they lack today has been supplied.
Will anybody act on the report?
The one part of the directive that could change behaviour has attracted almost no attention. The mediator’s report is not part of the trial papers. It is shown to the judge only on the question of costs, and only after the merits have been decided. Where the mediation has failed the mediator must record why, and the Protocol issued with the directive tells him what to note, whether a party failed to verify its mandate, failed to attend, came unprepared, sent someone without authority to settle or did not participate in good faith. That is a record of conduct, written by the only neutral person in the room, delivered to the judge at the moment costs are decided.
Whether a judge will act on it is a different question, and the answer is less comfortable than most practitioners assume, because the willingness of this Division to punish a party’s approach to mediation in costs is considerably older than the directive. In the case of MB v NB[2], decided in Johannesburg in 2009, Brassey AJ held that attorneys who had rejected mediation at a pre-trial conference in a matrimonial matter had failed their clients, and limited what they could recover from their own clients to the lower party and party scale. Rule 41A did not exist for another eleven years.
The same instinct runs through the matters now emerging under the directive. In the case of Oosthuizen[3], decided in Pretoria in March 2026, the applicants, whose mediation under the Protocol had failed, broke an undertaking about dates and left the other side four days, including a weekend, in which to answer. Minnaar AJ treated the application as an abuse of the urgent court and ordered punitive costs on the attorney and client scale. In the case of Shell Downstream[4], decided in Johannesburg in May 2026, Allen AJ held a rule 41A notice inadequate because it asserted that the disputes had no realistic prospect of settling without making out any case for that assertion.
What I have not yet found is a reported judgment in which the mediator’s report itself has moved a costs order. That is likely to be a question of time rather than principle, because most matters mediated under the directive have not yet reached trial. The safe assumption is not that the report will be ignored. It is that the report will be read in any matter where the other side has something to gain from drawing attention to it, and that is a large class of matters.
So however little you expect the mediation to achieve, treat it seriously. Attend the mediation properly, prepare properly and send someone with authority to settle. The sanctions for obstruction are separate from the report and have consequences of their own, from an order compelling attendance through to striking out and punitive costs, and for the practitioner personally, costs de bonis propriis, paid from his own pocket. None of that contradicts the point of this article. The directive can be enforced, and enforced sharply, and still not produce the settlements it was built to produce.
Is there a case for the directive?
There is, and it deserves to be put fairly. Trial dates running to 2031 are a denial of justice in everything but name. Gauteng carries roughly half the civil litigation in the country, its judicial complement was last increased in 2008, and a Judge President facing that has to do something procedural, because he cannot appoint judges.
The settlement on the morning of trial is also, from the court’s side of the bench, a waste. A judge has read the papers and prepared for a five day trial, a slot on the roll has been held for years, and both are consumed by a matter that was always going to settle while the case behind it in the queue waits another year. Seen from there, a process that resolves even a modest share of matters earlier is worth the inconvenience it causes to the rest. That is a policy judgment a Judge President is entitled to make, and I may be wrong about the proportion.
South Africa did not invent compulsory mediation. Italy, Singapore, Greece and parts of Canada and Australia all compel it in some form, and in several of those countries the court rolls did shorten. The honest caveat is that they funded their schemes and trained enough mediators before they compelled anyone.
The criticism that survives is not of the principle but of the timing. If the object is to capture the matters that will settle, the moment to compel mediation is when the pressure exists, which is in the weeks before trial rather than in the months after summons. A directive requiring mediation close to the trial date, when discovery is closed, the experts have reported and the client has begun to pay for the hearing, would do considerably more work than the present one, and it would cost the parties who are never going to settle a great deal less.
The matter will still end where these matters have always ended. Either the parties will meet on the morning, in the corridor where the silk collected his twenty-five years of experience, with everything finally real and the cost of the next five days sitting in front of them, and the matter will settle in the time it takes to draft the order. Or it will not settle, and a judge will decide it.
The directive has added a room. It has not moved the morning of the first day of trial.
Written by:
Elmien Liebenberg
September 2026
[1] Brondani v Brondani (2021/52977) [2025] ZAGPJHC 1157 (17 November 2025).
[2] MB v NB (2008/25274) [2009] ZAGPJHC 76; 2010 (3) SA 220 (GSJ) (25 August 2009).
[3] Oosthuizen NO and Others v Pistorius NO and Others (236657/2025) [2026] ZAGPPHC 219 (30 March 2026).
[4] Shell Downstream South Africa (Pty) Ltd v Shell Retailer Council (2023/132537) [2026] ZAGPJHC 576 (26 May 2026).
