A group of 64 Israeli twelfth-grade students of two different ethnic backgrounds participated in an experiment exploring the effects of argumentative design and social identity on the learning of a charged, ethnicity-related historical controversy. Students were divided into two learning conditions: an argumentative-disciplinary condition and a conventional textbook-based control condition. Students wrote short essays about Israel's "Melting Pot" policy of immigration absorption, prior to and after evaluation of historical sources and discussion. In the argumentative-disciplinary condition the final argumentative level of writing and the frequencies of stand and plot change were higher than in the control essays. As for confirmation bias, primary plot, stand, and argumentative level of pre-essays predicted final outcomes in the conventional textbook-based learning condition; no such relation existed in the argumentative-disciplinary condition. Narratives from the different ethnic groups differed in the frequency, direction, and degree of change, all toward improved in-group image. The design decisions toward the facilitation of argumentative activity seemed to facilitate narrative change, while social identity needs seemed to motivate it. (Contains 1 figure, 3 tables, and 10 footnotes.)