Jury instructions were constructed in a manner such that they differed only in terms of operative phrases and not with respect to linguistic or cognitive complexity. Two versions of a murder trial were used, and actors performed the trial on audiotape. Participants were told they would hear a trial and would then deliberate to a verdict. Results demonstrated that none of the jury instructions produced acceptable criteria of self-reported reasonable doubt, although instructions telling juries to be firmly convinced elicited the highest standards of proof. The firmly convinced instructions yielded verdicts that tended to correspond with the evidence in both versions of the trial. Firmly convinced juries focused proportionately more on the evidence and less on nonprobative issues than juries given other instructions. Juries had lower self-reported standards of proof when the trial evidence was weak for conviction as opposed to when it was strong. 23 references and 4 tables