The appellate courts - on more than one occasion - have indicated the legal standard for severing charges that are properly joined. The problem is, the standard so far exists only in theory, not in practice. The Oregon courts have never found a case that satisfy the express requirements for severance under current law (that I know of).
In Warren v. Baldwin, 140 Ore. App. 318, 324, 915 P.2d 1016, rev den, 324 Ore. 229, 925 P.2d 908 (1996), we explained that a trial court has discretion to deny severance if evidence of the joined offenses would be mutually admissible in separate trials, or if the evidence is sufficiently simple and distinct to mitigate the dangers created by joinder. The state concedes, and we agree, that, at the time of defendant's motion to sever, the evidence of the offenses would not be "mutually admissible." However, we also agree with the state that the evidence in the two cases is sufficiently simple and distinct that the trier of fact would have been able to separately consider the charges. See State v. McMinn, 145 Ore. App. 104, 107-08, 929 P.2d 1009 (1996). State v. Norkeveck, 214 Ore. App. 553, 560-561 (2007)
What kinds of facts would mean the evidence wasn't sufficiently simple and distinct to mitigate against the dangers of joinder? At the moment, there's no appellate example we can point to.
See also State v. McMinn, 145 Or App 104, 106 (1996), where the evidence was found to be "uncomplicated" and therefore permissibly joined.
The complexity of the evidence and the danger that the jurors will not be able to segregate the facts of one case from their determinations of the other case are relevant to the probable effectiveness of limiting instructions given to the jury by the court. State v. Staley, 142 Or App 583, 589, 923 P2d 650 (1996), rev den, 324 Or 560, 931 P2d 99 (1997).
Most people would say that the danger of joinder is the likelihood that the jury will use the multiple incidents as propensity evidence, a danger that could not be cured by lack of complexity, but it is what it is. Still, if you've messy, complicated cases that are properly joined for trial, this argument may be all that you've got.
Of course, if they aren't properly joined in the first place, you've got the improper joinder demurrer.