Most plaintiff lawyers treat a Code of Civil Procedure section 998 offer as a formality — a number typed into a template, served with the complaint, forgotten until the verdict comes in. That habit costs money. A well-built 998 is the only mechanism in California that lets you recover expert fees and add ten percent prejudgment interest to a personal-injury judgment, and it is the single most effective pressure a plaintiff can put on an adjuster who is sitting on reserves. A badly built one gets taxed to zero on a motion the defense files without breaking a sweat.
The rules have not changed much heading into 2026, but the way carriers litigate 998 validity has. Defense firms now challenge the reasonableness, the certainty, and the apportionment of nearly every offer that triggers a fee award. The offer you serve today has to survive a post-trial fight you will not see coming for two years. Below is how to write and time it so it does.
The Two Directions of the Shift
Section 998 runs both ways, and the asymmetry matters. When the plaintiff serves an offer, the defendant rejects it, and the plaintiff then obtains a judgment more favorable than the offer, subdivision (d) lets the court order the defendant to pay the plaintiff's post-offer expert witness fees. Pair that with Civil Code section 3291 and the plaintiff also gets prejudgment interest at ten percent per year running from the date of the offer on the personal-injury damages. In a case with two treating experts, a biomechanic, and an economist, that fee recovery alone can run into six figures, and the interest on a case that takes three years to try can dwarf the fee award.
Running the other direction, a defense 998 that the plaintiff rejects and fails to beat flips the plaintiff's own cost recovery. Under subdivision (c)(1) the plaintiff loses post-offer costs and can be ordered to pay the defendant's post-offer costs, including — at the court's discretion — the defendant's expert fees. That is the scenario that turns a modest plaintiff's verdict into a net loss for the client. The exposure is real enough that a lowball defense 998 has to be evaluated as seriously as any demand, because the downside of guessing wrong is charged against the recovery.
The Comparison Math People Get Wrong
The phrase "more favorable judgment" is not measured against the bare verdict. For a defense offer, the court adds the plaintiff's pre-offer costs to the verdict and compares that total to the offer amount. So a $95,000 verdict on a rejected $100,000 defense offer can still beat the offer if the plaintiff had $8,000 in recoverable pre-offer costs on the books at the time the offer was served. The date of the offer is the line that separates pre- from post-offer costs, which is why serving early is not only about pressure — it shrinks the pool of costs that get moved to the wrong side of the ledger if the case goes sideways.
Interest compounds the arithmetic in the plaintiff's favor. Because section 3291 interest runs from the date of a plaintiff's 998 offer rather than from judgment, an offer served ten days after the answer, in a case that reaches verdict three years later, carries three full years of ten-percent interest if the plaintiff beats it. On a $600,000 verdict that is $180,000 the defense cannot touch, added to a fee award. The instinct to hold the offer until you have "developed the case" is usually backwards. Serve a defensible number early and let the interest clock do the work.
Building an Offer That Survives
The two grounds that kill 998 offers are unreasonableness and uncertainty, and both are avoidable. An offer must be made in good faith and be reasonable when served, judged on what the offeror knew at the time — a token offer with no realistic prospect of acceptance, made only to trigger fee-shifting, is not valid. The factors trace back to Elrod v. Oregon Cummins Diesel, Inc. (1987) 195 Cal.App.3d 692: whether the offer was within the range of reasonably possible outcomes, and whether the offeree had access to the information needed to evaluate it. The practical takeaway is that a plaintiff's offer served before the defense has had any discovery is vulnerable, because the defendant can argue it could not fairly assess the number. Serve after the defense has your medical records and your demand package, not before.
Certainty is the other trap. The offer must state terms clear enough that the offeree can accept with a signature and know exactly what it bought. Avoid conditions that require future negotiation, dueling releases, or indemnity language the defense can characterize as an unaccepted new term. A clean 998 states a dollar figure, provides for entry of judgment or a dismissal with a defined release, and specifies who bears costs. The moment you bolt on a broad hold-harmless clause or a confidentiality provision with teeth, you have handed the defense a certainty argument for the motion to tax.
The Joint-Offer and Multiple-Party Problem
Cases with several defendants or several plaintiffs are where valid offers go to die. A joint offer made to multiple defendants, or by multiple plaintiffs, is generally invalid unless it is apportioned among them and each party can independently evaluate and accept its share — the reasoning developed in cases like Peterson v. John Crane, Inc. (2007) 154 Cal.App.4th 498. An unapportioned lump offer to three defendants forces each to weigh its exposure against a number it cannot accept alone, and courts treat that as no valid offer at all. The fix is mechanical: serve a separate, individually numbered 998 on each defendant, each capable of standing on its own. It is more paper, but it is the difference between a fee award and a taxed motion.
The same discipline applies when your case sits inside a larger fact pattern with allocation questions. Comparative-fault exposure changes the reasonableness analysis, and the way juries have been splitting responsibility lately should inform the number you put on the offer — the trends we tracked in where California juries are putting the fault line on plaintiffs are the same trends a defense expert will cite when arguing your offer was never reasonable. In coverage-driven cases, the identity of the paying party matters too; disputes like the one in Lanunziata v. Penn National and the residency line in UIM can determine whether there is even a solvent target for a 998 to bind.
Post-Trial Enforcement and the Motion to Tax
Winning the verdict is not winning the 998. Expert fees under subdivisions (c) and (d) are discretionary, and the court decides them on a motion after judgment. The plaintiff who wants those fees has to document them contemporaneously: dated invoices, hours, hourly rates, and a clear line between pre-offer and post-offer work, because only post-offer expert fees are recoverable. Reconstructing that record after the verdict is how good fee claims get cut in half. Keep the expert billing clean from the day the offer is served.
Expect the defense to attack on every front the statute allows — that the offer was unreasonable when made, that it was uncertain, that it was not properly served, that the judgment did not actually exceed it once the correct costs are counted. The interest question under section 3291 is a separate ruling, and it is worth briefing on its own rather than folding it into the cost memorandum. Enforcement discipline is the same discipline that decides other post-recovery fights, from hospital lien perfection defects to costs and interest — the money is in the paperwork, and the side with the cleaner record usually keeps it.
Where This Leaves You
The 998 is not a settlement tactic you deploy at mediation. It is a document you draft the way you draft a complaint — early, defensibly, and with the post-trial fight already in view. Serve a number a court will call reasonable, keep it certain, apportion it when there is more than one party on either side, and bill your experts so the record proves itself two years later. Done that way, a rejected 998 is not a missed settlement. It is a lien on the verdict the defense chose to make bigger.