On May 24, 1978, six year old Jose DeJesus fell over the porch railing of a third-floor apartment on Broadway in Cambridge, sustaining substantial injuries. His mother had taken him there to visit a friend who was a tenant of the defendant, who owned the building.
This action, based on a claim of negligence of the landowner, was submitted to a jury on special questions. They answered that the defendant was negligent but that her negligence was not a proximate cause of injury to Jose.
On Jose’s appeal, the Appeals Court (in an unpublished memorandum and order under Appeals Court Rule 1:28) ordered a new trial, concluding that the trial judge improperly excluded as too remote testimony concerning the looseness of the porch railing approximately two months before Jose’s fall. The defendant, seeking leave for further appellate review, argued to us that the evidentiary error, if it was one, could not have prejudiced the plaintiff because the excluded evidence related exclusively to the issue of the defendant’s negligence, a point on which the jury found for the plaintiff.2 We granted further appellate review to consider in part whether the judge was warranted in excluding the evidence as too remote and principally whether a new trial need not be ordered, in any event, because any error in excluding the evidence could not have been prejudicial to the plaintiff.
*46We agree with the Appeals Court that the judge erred in excluding relevant evidence that, as a matter of law, was not remote in time to the accident. We further conclude that the error was prejudicial, in the sense in which that word is properly applied to decide at the appellate level whether trial error is reversible error. Consequently, as the Appeals Court concluded, there must be a new trial.
The railing on the porch of the third-floor apartment was approximately thirty to thirty-one inches above the porch floor. There was an horizontal bottom section of the railing, the top of which was seven and one-half inches above the porch floor. Jose was fifty inches tall. There was evidence that Jose put his feet between the vertical rails of the porch railing, stood on the bottom section of the railing, and looked down at some youngsters. He then fell. There was some evidence, some of it conflicting, concerning the condition of the railing. The plaintiff presented testimony that after the accident the railing was loose. There was testimony in the course of the presentation of the plaintiff’s case, from the defendant’s son-in-law (the property’s manager) and from a carpenter employed by the defendant, that the railing was solid after the accident and that no changes had been made in the porch for a substantial period before the accident.
The plaintiff subsequently offered the testimony of Ana Maldonado, a tenant in the second-floor apartment, concerning the state of the porch railing when she visited the third-floor apartment about two months before the accident. The judge excluded that testimony on the ground that it was remote in time. The plaintiff by offer of proof indicated that Maldonado would testify “that the porch rail in question was loose approximately two months before the date of the accident.”
1. We agree with the Appeals Court that Maldonado’s testimony concerning the condition of the railing about two months before the accident was improperly excluded. The evidence, as the defendant concedes, was relevant. See Tosti v. Ayik, 394 Mass. 482, 489-490 (1985), S.C., 400 Mass. 224, cert. denied sub nom. UAW Local 442 v. Tosti, 484 U.S. 964 (1987), citing Green v. Richmond, 369 Mass. 47, 59 (1975); Poirier v. *47Plymouth, 374 Mass. 206, 210 (1978). Hers was the only evidence concerning the looseness of the railing before the accident and thus was not duplicative of other testimony. Witnesses favorable to the defendant had testified that there had been no changes on the porch for more than two months before the accident.
The general pattern of our cases on the alleged remoteness in time or space of particular evidence indicates two general principles. If the evidence had some probative value, decisions to admit the evidence and to leave its weight to the jury have been sustained. See Poirier v. Plymouth, supra; Sherburne v. Meade, 303 Mass. 356, 362 (1939); Nelson v. Old Colony St. Ry., 208 Mass. 159, 161 (1911). The exclusion on the ground of remoteness of relevant evidence has generally not been sustained. See Crowe v. Ward, 363 Mass. 85, 88-89 (1973). (“We are influenced by the general view that relevant evidence should be admitted unless there is a quite satisfactory reason for excluding it”); Kramer v. John Hancock Mut. Life Ins. Co., 336 Mass. 465, 468 (1957) (relationship of excluded evidence to the issue in the case would not have been “mere conjecture,” and it should have been admitted).3 The cases have recognized a range of discretion in the judge. See Crowe v. Ward, supra. Where there is evidence offered that a porch railing was loose two months before an accident and persons who would have acted for the landlady in ordering and making repairs have testified that there were no repairs made during this period, the offered evidence was relevant, not remote, and should have been admitted.
2. If, as we have concluded, Maldonado’s testimony of pre-accident looseness of the railing should have been admitted, we must consider whether the plaintiff has demonstrated prejudice that entitles him to a new trial. The Legislature has said that an error in the exclusion of evidence should not be grounds for a new trial unless the error has “injuriously affected the *48substantial rights of the parties.” G. L. c. 231, §§ 119 and 132 (1986 ed.).4 As will be seen from our discussion, the substantial rights of a party are adversely affected when relevant evidence is erroneously excluded that, viewing the record in a commonsense way, could have made a material difference.
Application of the concept of prejudicial error to the erroneous exclusion of evidence requires a case by case analysis of the effect of the error. Surely, the concept cannot place on the allegedly aggrieved party the burden of demonstrating that the trier of fact would certainly have come to a different conclusion had the evidence been before it. In Crowe v. Ward, supra at 90, we said, ordering a new trial, that erroneously excluded evidence “could have materially affected the result.”5 We have also said that the exclusion of particular evidence would not require a new trial where, analyzing the facts, it was most unlikely that the jury would have been affected by the error (if indeed there had been any error). Drake v. Goodman, 386 Mass. 88, 94 (1982).6 See Daley v. American Printing Co., 150 Mass. 77, 81 (1889) (the erroneous exclusion of evidence requires a new trial “unless it can be seen that, even if it had been admitted and believed, still the verdict” would have been the same). But see Bendett v. Bendett, 315 Mass. 59, 65 (1943) (party showing error in exclusion of evidence must also show “reasonable probability of harm”).
We think the appropriate test is whether the proponent of erroneously excluded, relevant evidence has made a plausible *49showing that the trier of fact might have reached a different result if the evidence had been before it. Thus the erroneous exclusion of relevant evidence is reversible error unless, on the record, the appellate court can say with substantial confidence that the error would not have made a material difference.
3. We come, therefore, to the question whether, if the erroneously excluded evidence had been admitted, it might have made a material difference in the jury’s special verdict. The defendant’s argument that the excluded evidence would have made no difference has a superficial appeal. Maldonado’s testimony concerned a looseness of the porch railing observed before the accident. Because there was no other direct evidence of pre-accident looseness, it cannot fairly be said that the evidence was cumulative and hence its exclusion not prejudicial.7 It can, however,, be argued that Maldonado’s testimony concerned only the defendant’s alleged negligence, on which the jury found for the plaintiff, and thus the plaintiff could not have been prejudiced by the error.
If evidence of the looseness of the railing (based on witnesses’ postaccident observations) had been the only evidence of negligence that could have caused the child’s injuries, the defendant would be correct in her argument that there was no prejudicial error. The jury’s special verdict would have stated in effect that the railing was negligently loose but that negligence was not the proximate cause of the child’s fall. There was, however, evidence that wood on the railing was rotted and that the railing bowed outward. This evidence might have been the basis of the jury’s conclusion that the defendant was negligent but that none of these circumstances caused the child to fall. There was also evidence that the top of the railing was about thirty inches off the floor and that the lower horizontal portion of the railing, on which a child could stand, was seven and one-half inches above the floor. This evidence also could *50have supported a finding of negligence, but a finding that the railing’s design did not cause the child to fall would also have been warranted.
Evidence directly to the effect that the railing was loose sometime before the accident might have led the jury to find that it was loose at the time of the fall, although without that evidence the jury in fact made no such finding. We think the jury, as a practical matter, might have come to a different conclusion on causation if it had found that the defendant negligently permitted the railing to be loose. We cannot say with substantial confidence that admission of the evidence that the railing was loose would not have made a material difference in the jury’s special verdict. A bowed, rotted, low railing with a lower horizontal rail that is seven and one-half inches off the floor might not have caused the child to fall. If the jury had found that same railing was also loose and shaky, perhaps they would have further found that the defendant’s negligence was the proximate cause of the child’s fall. We think the plaintiff has made a plausible showing that the jury might have reached a different result if the excluded evidence had been before it. We should not usurp the jury’s function.
4. We agree with the Appeals Court that there must be a new trial.
Judgment of the Superior Court reversed.
The defendant’s brief in the Appeals Court did not make the argument that any error in excluding the evidence was not prejudicial. The plaintiff did not file an opposition to further appellate review pointing out that omission. As an argument in support of judgment entered in the trial court, an appellate court may properly consider it, although it need not do so when it is raised for the first time in an application for further appellate review. See McLeod’s Case, 389 Mass. 431, 434 (1983); Amherst Nursing Home, Inc. v. Commonwealth, 398 Mass. 850, 852 (1986).
The application for further appellate review states that the plaintiff waived three issues in the Appeals Court and that only one issue remained for the Appeals Court to consider. In his supplemental brief here, however, the plaintiff presses another argument. Because there is to be a new trial, we need not discuss it.
Our older cases sustaining the exclusion of evidence on the ground that it was “too remote” may fairly be read as saying that, in the circumstances, the evidence was not relevant. See Bachinsky v. Rogers, 273 Mass. 381, 385 (1930); Commonwealth v. Abbott, 130 Mass. 472, 474-475 (1881).
Likewise, Mass. R. Civ. P. 61, 365 Mass. 829 (1974), advises the trial court that an error in the exclusion of evidence should not be the basis for relief unless the substantial rights of a party have been affected.
A weather report relevant to contested issues of fact in the case was improperly excluded. We concluded that the error could have materially affected the jury verdict for the defendant. Id.
In the Drake case, we observed that the exclusion of evidence dealing solely with whether a physical condition was caused by the defendant’s allegedly negligent conduct was not reversible error, because on the record it was most likely that the jury’s verdict for the defendant was based on a failure of proof of negligence. If there had been a finding of negligence, the jury, as a practical matter, would have returned a verdict for the plaintiff in some amount.
In some instances, an error would not prejudice the case because the improperly excluded evidence would have been “merely cumulative” of other evidence pointing in the same factual direction. See Pina v. McGill Dev. Corp., 388 Mass. 159, 164 (1983); Abele v. Dietz, 312 Mass. 685, 690-691 (1942).