A total of 237 students enrolled in one of several undergraduate courses at a large, urban, midwestern university participated in the study. Participants were scheduled as groups of 7-10 and were assigned to mock juries that consisted of four to seven individuals. There were 20 groups with 4 participants, 19 groups with 5 participants, 8 groups with 6 participants, and 2 groups with 7 participants. Data were collected on 49 groups. The mock juries read an edited transcript from a real murder trial and deliberated to a unanimous verdict. Juries with the option to convict on a lesser charge produced more overall convictions than juries that received only the primary charge, but only when "reasonable doubt" was undefined. Conversely, more murder convictions resulted when the "reasonable doubt" criterion was defined in relation to being "firmly convinced" of the defendant's guilt. Parallel analyses of the effects of these two procedural factors on individual juror predeliberation verdict preferences differed in several important ways, highlighting the need for group-level research on juries. Implications, limitations, and a preliminary integrated model of jury decision making are discussed. 2 tables, 2 figures, and 34 references