Getting “prior accidents” admitted in cases against public entities

The substantial similarity rule, its limits, and how to counter the defense’s “absence of accidents” argument

2026 August

In a dangerous-condition-of-public-property case, prior accidents at the same location can be among the most powerful evidence a plaintiff presents. Prior accidents prove the public entity had notice of the hazard. They can help prove the condition was dangerous. And they shape the way the jury understands causation, by establishing that what happened to the plaintiff was not an aberrant event, but the predictable consequence of a known hazard.

Prior-accident evidence does not come in automatically. The moving party must show substantial similarity between the subject accident and any prior accidents, and the rule applies different levels of stringency depending on what the evidence is offered to prove. This article walks through the framework, its limits, and the counter to the defense argument that the absence of similar accidents proves a location is safe.

Why prior accidents matter under Government Code § 835

A claim for dangerous condition of public property requires proof that the property was in a dangerous condition at the time of the injury, that the injury was proximately caused by that condition, that the condition created a reasonably foreseeable risk of the kind of injury that occurred, and that the public entity either created the condition or had actual or constructive notice of it. (Gov. Code, § 835.)

Prior-accident evidence speaks directly to several of these elements. 

Notice

A public entity has actual notice when it knows of the condition’s existence and dangerous character (Gov. Code, § 835.2, subd. (a)), and constructive notice when the condition was of such an obvious nature that it should have been discovered (Gov. Code, § 835.2, subd. (b)). 

Existence of a dangerous condition

A pattern of prior accidents involving the same hazard is direct evidence that the property creates a substantial risk of injury. 

Foreseeability

The entity’s own records of prior accidents establish that the risk was foreseeable. Building this record is therefore a priority in any public entity case.

The substantial-similarity rule

“As a general rule, ‘[w]here the circumstances are similar, and the happenings are not too remote in time, other accidents may be proved to show a defective or dangerous condition, knowledge or notice thereof, or to establish the cause of an accident. . . .’ The evidence must relate to accidents which are similar and which occur under substantially the same circumstances.’” (Genrich v. State of California (1988) 202 Cal.App.3d 221, 227–228, italics in original.) 

The substantial-similarity doctrine does not operate by formula. The trial court applies the framework to the particular facts of the case, and the standard’s stringency varies according to the purpose for which the evidence is offered. The flexibility cuts both ways, giving plaintiff’s counsel room to argue for admission of evidence that is similar in the relevant respects even when not identical, and giving the court room to admit substantial groups of evidence on notice that would be excluded under the stricter standard for dangerous condition. 

The judge is the gatekeeper

The admissibility of prior-accident evidence is committed to the sound discretion of the trial court. (Simmons v. Southern Pac. Transportation Co. (1976) 62 Cal.App.3d 341, 365.) The substantial-similarity doctrine is not a bright-line test that admits or excludes evidence in some formulaic fashion. It is a framework the court applies to the particular facts before it, with broad latitude to weigh probative value against the risks of confusion of issues, undue consumption of time, and unfair prejudice.

The gatekeeping function has several practical consequences for plaintiff’s counsel.

First, it is important to win the battle of defining the hazard mechanism for substantial similarity purposes. If counsel frames the hazard as the geometric or operational feature that caused the injury, broadly enough to capture the universe of similar prior accidents but narrowly enough to exclude obvious outliers, the court can admit a substantial amount of evidence. If the defense succeeds in narrowing the frame to the precise circumstances of the subject incident, very little will fit. Plaintiff’s counsel should make the affirmative framing in motion in limine briefing, with expert declarations identifying the mechanism and a substantial similarity analysis for each prior accident offered.

Second, trial courts that admit prior-accident evidence are rarely reversed. Trial courts that exclude on substantial-similarity grounds are also generally affirmed. The discretionary nature of the determination means that the trial court’s ruling is afforded great weight, which makes the pretrial record the place where these issues are typically decided.

Third, the gatekeeping role applies to all parties. When the defense offers purported safety evidence, aggregated total-collision rates, or other comparison data to show that a location is safe, the trial court has the same discretion to exclude under Evidence Code section 352 that it applies to the plaintiff’s evidence. A defendant’s comparison data that aggregates non-similar accidents is no more probative of safety than the plaintiff’s non-similar accidents would be of dangerousness.

A stricter standard for proving dangerous condition

The substantial-similarity standard is not uniform. It varies in stringency depending on the issue to which the evidence is directed: “[w]hile there must be substantial similarity to offer other accident evidence for any purpose, a stricter degree of substantial similarity is required when other accident evidence is offered to show a dangerous condition; ‘the other accident must be connected in some way with that thing.’” (Salas v. Dept. of Transportation (2011) 198 Cal.App.4th 1058, 1072.) 

This is the underlying principle: “The requirement of similarity may vary in strictness according to the purpose for which the evidence is introduced. Thus, if offered to show a dangerous condition of a particular thing – such as a step – the other accident must be connected in some way with that thing.” (Genrich, supra, 202 Cal.App.3d at 227-228.)

The requirement of the connection to “that thing” is the crux of substantial similarity for the purposes of showing dangerous condition. A prior accident is substantially similar when it arises from the same hazard mechanism. This would include the same physical defect or configuration operating on a similarly situated user. A different mechanism, even at the same location, does not satisfy the rule when offered to prove the existence of the dangerous condition alleged.

The case law is illustrative. In Salas, the plaintiff was a pedestrian struck and killed while crossing a state highway. The plaintiff offered 23 earlier vehicle-to-vehicle collisions at the same location. The court rejected them as not substantially similar because “none of the proffered accidents even involved a pedestrian, much less a pedestrian who stopped while crossing the street and then changed direction.” (Salas, supra, 198 Cal.App.4th at 1072.) In Mixon v. State (2012) 207 Cal.App.4th 124, 138, prior accidents involving different lighting conditions, different positions in the intersection, and a drunken pedestrian were held not substantially similar. In Sambrano v. City of San Diego (2001) 94 Cal.App.4th 225, 238-239, a fire-ring incident in which a child tripped and fell into a smoldering ring was held not substantially similar to a child climbing into a different, much larger fire ring, because the mechanisms of injury differed. 

The pattern in the case law is clear. Courts look for a true causal connection between the prior accidents and the specific hazard alleged. Generalized “accidents in the vicinity” do not suffice when the evidence is offered on the dangerous-condition element.

At the same time, the substantial similarity analysis does not focus on every surface-level detail of how the accidents unfolded. Arguable differences that do not change the underlying mechanism go to weight, not admissibility. Johnston v. County of Yolo (1969) 274 Cal.App.2d 46, illustrates the point. There, the plaintiff was injured when a southbound vehicle failed to negotiate a sharp double curve. The trial court admitted evidence of 10 earlier accidents at the same double curve, all involving northbound vehicles. The county argued on appeal that the prior accidents were dissimilar because southbound traffic would slow for a narrow bridge before reaching the curve while northbound traffic had no such cue. The Court of Appeal affirmed admission: “The double curve, however, had the same geometry whether approached from north or south. Differences in the approaches could be argued to reasonable jurors and understood by them.” (Id. at 59.) In that case, the same hazard, the geometry of the curve, operated on both classes of users. The directional difference was for the jury to weigh, not for the court to use to exclude.

The much-relaxed standard for notice

The substantial-similarity rule loosens considerably when the evidence is offered to prove notice rather than the existence of a dangerous condition. In Hilts v. County of Solano (1968) 265 Cal.App.2d 161, a witness testified to “at least a dozen accidents” at the subject intersection, although he had not actually witnessed any of them, was unsure how many there were, and acknowledged intoxication might have played a part in some. The County argued the testimony was inadmissible for lack of substantial similarity. The Court of Appeal rejected the argument: “The requirement of similarity of conditions is ‘much relaxed’ when the evidence is offered to show notice of the dangerous condition.… The test of admissibility of evidence of prior accidents to show notice is that the evidence must be such as to have attracted the defendant’s attention to the dangerous situation.” (Hilts, supra, 265 Cal.App.2d at 169.)

The purpose of notice evidence is to show what the public entity knew about a problem location, including what reports it received, what complaints it logged, and what its own records reflected. The probative value does not depend on whether each earlier incident was identical in every respect, but on whether, taken together, the information was enough to alert a reasonably careful public entity to investigate. Genrich reaffirms the same approach, permitting lay testimony from a neighborhood resident about “screeching brakes,” “burning rubber,” and “near misses” at the subject intersection, along with his calls to Caltrans, all for the limited purpose of showing notice. (Genrich, supra, 202 Cal.App.3d at 231-233.)

In one recent case involving a broadside collision at an intersection, we successfully argued that 34 prior broadside collisions at the same intersection were admissible to prove notice, while 32 were admissible to prove the dangerous condition. The court ruled that the broader category of broadside collisions, in which vehicles entering from a side street are struck by through traffic, put the public entity on notice that the intersection was hazardous because left-turning motorists were misjudging traffic gaps. Two of the prior accidents did not fit the precise left-turn-across-path mechanism that defined the subject incident and so they were not admissible on the issue of dangerous condition, but the broader issue of notice encompassed them. 

The same prior accidents can serve multiple purposes at once

A point that is sometimes lost in the dangerous-condition-versus-notice framing is that the same prior-accident evidence can serve both purposes simultaneously, along with causation. Magnuson v. City of Stockton (1931) 116 Cal.App. 532 is directly on point. In Magnuson, the plaintiff offered evidence that three other children had drowned in the same lake before her son drowned there. The trial court admitted the evidence but instructed the jury that it could be considered only for the limited purpose of showing notice. The Court of Appeal reversed, holding that the instruction was error, and the evidence was properly admitted for three purposes: to show the dangerous condition existing at the lake, to show the cause of the boy’s death, and to show the entity’s knowledge of the dangerous condition. (Id. at 535.)

The lesson is important. Public-entity defendants routinely propose limiting instructions confining prior-accident evidence to notice only. Where the prior accidents also satisfy the stricter dangerous-condition standard, plaintiff’s counsel should resist the narrow instruction proposed by the defense and request one that reflects all the purposes for which the evidence is admissible. An instruction that artificially restricts the jury’s use of the evidence to one element is reversible error. 

Prior claims and lawsuits against the entity are themselves notice

At times, the underlying accident may result in the filing of a government claim or even a lawsuit against the public entity. Prior government claims and prior lawsuits arising from the same condition are competent evidence of notice, independent of the underlying accidents they describe. Bigelow v. City of Ontario (1940) 36 Cal.App.2d 198, is the foundational authority. There, the plaintiff offered a government claim previously filed against the city by another injured motorist at the same hump in the roadway. The trial court refused to admit the claim or even allow it to be marked for identification. The Court of Appeal reversed, calling the ruling “serious error” and observing that “there could be no more competent evidence of notice of a dangerous or defective condition of a street than a claim filed by a person who had been injured while traveling over it.” (Id. at 203-204.)

The same logic applies to prior lawsuits. Bigelow held that an earlier suit against the city for injuries at the same location “was notice of the condition and should have put the city upon its inquiry,” and that the prior suit’s unsuccessful outcome did not change the analysis because “[t]he question of whether or not that condition was dangerous was a fact to be determined in this action and did not depend on the outcome of the suit.” (Id. at 204.) A prior claim or suit gives the entity formal, documented notice of the condition, regardless of how the prior matter was ultimately resolved. The practical implication is straightforward. In every dangerous-condition case, request the entity’s claims log and litigation history relating to the subject location through targeted CPRA requests and written discovery.

The important limits

The relaxed-notice standard is generous, but it also has its limits. 

Temporal: Subsequent accidents do not prove notice

A subsequent accident can be admissible to show that the condition was dangerous, but it is not relevant to notice. A defendant cannot have been on notice of a danger based on an event that had not yet occurred. The Genrich court was explicit: “[A] subsequent accident, though admissible to show a condition was dangerous, is not relevant on the issue of knowledge or notice of a dangerous condition existing at the time of the injury.” (Genrich, supra, 202 Cal.App.3d at 228-229.) Therefore, a plaintiff seeking to use post-injury accidents must do so only to prove dangerous condition, and only after satisfying the stricter substantial similarity standard that purpose requires. 

Evidentiary: How the data comes in matters

Prior-accident data is often documentary, based upon collision databases, police reports, and internal entity records. Much of this material presents hearsay and foundational issues if offered for its truth. 

Under Evidence Code section 801, subdivision (b), an expert may rely on otherwise-inadmissible material if it is of a type reasonably relied on by experts in the field. Under Evidence Code section 802, the expert may state on direct examination the matter on which the opinion is based. Traffic-safety experts routinely rely on traffic-collision reports and SWITRS data, peer-reviewed safety studies, and the entity’s own internal records when forming opinions about dangerous condition and notice. Genrich confirms that an expert may identify such material and describe it as the basis for the opinion. (Genrich, supra, 202 Cal.App.3d at 229.)

The limit is that the expert may not use the basis-for-opinion principle to recite hearsay details for the truth of the matter asserted. (Grimshaw v. Ford Motor Co. (1981) 119 Cal.App.3d 757, 788-789.) The conventional safeguard is a limiting instruction stating that the matter on which the expert relied is admitted only to show the basis of the opinion, not for its truth. After People v. Sanchez (2016) 63 Cal.4th 665, the lines around expert-based case-specific hearsay have sharpened. An expert may testify to background knowledge but cannot relate as true case-specific facts asserted in inadmissible hearsay unless those facts are independently proven or covered by a hearsay exception. Whenever possible, work to get the underlying records in directly, whether through a custodian of records, the entity’s person most knowledgeable, judicial notice, or admissions, rather than relying solely on the expert to introduce them.

A procedural trap: Preserve the objection

Hilts also teaches a procedural lesson. There, the County objected to prior-accident testimony for lack of substantial similarity. The trial court overruled the objection, but invited cross-examination on similarity. The County developed the lack of similarity on cross, and then failed to move to strike. The Court of Appeal held the objection waived. (Hilts, supra, 265 Cal.App.2d at 168-169.) The same trap operates against plaintiff’s counsel when the entity offers comparison evidence. Once cross-examination establishes a deficiency, move to strike on the record.

The defense’s “absence of accidents” argument, and how to counter it

Public-entity defendants regularly offer the absence of similar prior accidents as affirmative evidence that a condition is safe. The argument has appellate support. In Sambrano, the Court of Appeal expressly held that the city’s “safety history,” including the lack of similar accidents over five years of heavy park use, was relevant to the issue of dangerous condition. (Sambrano, supra, 94 Cal.App.4th at 243.) Public entities attempt to offer this kind of evidence at summary judgment and at trial. There are four lines of authority and argument that counter the move.

1. Absence of accidents is not dispositive

The absence of prior accidents is, at most, one factor among many. It does not establish non-dangerousness as a matter of law. (Lane v. City of Sacramento (2010) 183 Cal.App.4th 1337, 1346.) A condition may be dangerous even if no one has yet been injured. The question is whether the property creates a substantial risk of injury when used with due care, informed by, but not controlled by, the historical accident record.

2. A known physical deficiency does not become unknown for lack of accidents

When the dangerous condition is a physical or geometric feature, such as a sight-distance restriction, a roadway configuration, or a missing safety device, the entity’s knowledge of the physical condition controls the notice inquiry. The Court of Appeal made this point directly in Cole v. Town of Los Gatos (2012) 205 Cal.App.4th 749, 779-780, reversing summary judgment because triable issues of notice existed where the town knew the physical characteristics that created the danger, even absent prior accidents. In Carson v. Facilities Development Co. (1984) 36 Cal.3d 830, 843, the court held that whether a dangerous condition could have been discovered by reasonable inspection, and whether there was adequate time to take preventive measures, are properly left to the jury.

The argument is especially powerful in cases involving objective engineering deficiencies. When the entity’s own standards specify a minimum value and the location does not meet that value, the entity is on notice of the deficiency by virtue of knowing the physical layout, regardless of whether anyone has yet collided there.

3. The defense’s comparison data must satisfy substantial similarity too

Substantial similarity is a rule of relevance. Thus, if a plaintiff cannot offer non-similar prior accidents to prove that a condition is dangerous, a defendant cannot offer non-similar prior accidents, or their absence, to prove that a condition is safe. The rule rests on Evidence Code section 350, in which only relevant evidence is admissible, and relevance requires a logical connection to the issue in dispute. A total collision figure or “safety history” that aggregates the specific hazard at issue with rear-ends, sideswipes, single-vehicle run-offs, drunk-driving collisions, and other unrelated accident types is not relevant, as to either party, on the issue of whether the specific hazard makes the location dangerous.

Caltrans’s showing in Salas relied in part on the absence of pedestrian-vehicle collisions over a 10-year, 31,000,000-vehicle period. The court accepted that showing, but only because Caltrans’s data was tailored to the same accident type as the alleged condition. Caltrans was not permitted to prevail on “no prior collisions of any kind.” It prevailed because its comparison evidence matched the hazard. (Salas, supra, 198 Cal.App.4th at 1071-1072.) The same rule applies to both sides.

When the entity offers aggregated rate data or evidence of safety history not targeted to the alleged hazard, consider a motion in limine arguing that the data fails substantial similarity in the same way the plaintiff’s dissimilar prior accidents would.

4. No inspection system means no defense to constructive notice

Constructive notice does not require a showing of prior accidents. Under Government Code section 835.2, subdivision (b), constructive notice may be shown by establishing that the condition existed for a period of time and was of such an obvious nature that the entity, in the exercise of due care, should have discovered it, or, alternatively, that a reasonable inspection system would have discovered it. (See also Straughter v. State of California (1976) 89 Cal.App.3d 102, 109; Erfurt v. State of California (1983) 141 Cal.App.3d 837, 844-845.) Where the entity admits that it has no inspection system for the kind of hazard at issue and relies on public complaints to surface problems, the constructive notice analysis breaks in the plaintiff’s favor.

Bolster this with expert testimony that the deficiency would have been apparent to any competent engineer or maintenance employee at the location. The absence of accidents then tells a different story. Not that the location was safe, but that the entity was lucky for a period of time while it failed to inspect a known hazard.

Practical steps for counsel

A few things to do in every dangerous condition case against a public entity.

Search SWITRS first. The Statewide Integrated Traffic Records System, maintained by the California Highway Patrol, is publicly available and searchable, and it captures collisions reported throughout California, regardless of which entity owns the roadway. Pull the SWITRS history for the subject location and surrounding area. SWITRS may not capture every collision, and the entity’s own records will be more complete (e.g., Caltrans’s TASAS database is more comprehensive, but requires a public records request), but SWITRS gives counsel an early peek at the accident pattern.

Submit broad public records requests early. CPRA requests should seek the entity’s collision history database extracts (such as TASAS for Caltrans matters), individual traffic collision reports, traffic studies, complaint logs, internal safety analyses, and inspection records and protocols. Include the entity’s government claims log and litigation history for the subject location. 

Identify the hazard mechanism early. The substantial similarity analysis turns on identifying the specific feature that created the risk. Work with the expert to define the mechanism as precisely as possible. Once the hazard is defined, both the universe of substantially similar prior accidents and the universe of objective engineering deficiencies become identifiable.

Develop the connection in expert testimony. The plaintiff’s expert should be prepared to testify, for each prior accident offered, why it shares the same hazard mechanism as the subject incident. Prepare a limiting instruction directing the jury how to consider the underlying records the expert relied on.

Separate prior from subsequent accidents. Pre-injury accidents can be used as evidence for both notice and dangerous condition, while post-injury accidents can prove only dangerous condition. 

Anticipate the defense based upon the lack of accident history. Identify the entity’s showing regarding safety history and arguments regarding the lack of accidents. Develop the Cole/Carson arguments discussed above and the lack of an adequate inspection system with both the expert and the entity’s person most knowledgeable.

Preserve the record. Object on substantial similarity, develop the deficiency on cross, and move to strike. An unstricken answer is in evidence under Hilts.

Conclusion

Prior-accident evidence is typically of primary importance in cases for dangerous condition of public property. The law that controls its admission is workable, but it has nuances that may catch the inattentive practitioner. Distinguish carefully between the dangerous condition and notice purposes for which the evidence is offered. Watch the temporal line between prior and subsequent accidents. And when the entity invokes its lack of accident history defense, hold it to the same substantial similarity standard the plaintiff has to meet. Develop the parallel theory: A known physical deficiency, in a system without inspections, establishes notice no matter how lucky the entity has been.

Scott E. Boyer is an attorney at The Homampour Law Firm in Los Angeles, where he represents individuals in catastrophic injury and wrongful-death matters, often against public-entity defendants. He received his B.A. from U.C. Santa Barbara and his J.D. from U.C. Davis School of Law.