The present study investigates conditions under which undergraduates may adopt a view of history more similar to that of historians and how learning and understanding may be affected under such conditions. Two manipulations, one of the reading material and the other of writing task, were introduced within the standard ‘read-to-write’ approach of history instruction. Undergraduates were either given a textbook chapter about Ireland between 1800 and 1850, or the same information in the form of separate sources. After reading the presented material they were instructed to write a history, a narrative or an argument regarding what produced the significant changes in Ireland's population between 1846 and 1850. It was expected that the separate source/argument writing condition would yield the most historian-like behaviour. Indeed, students in this condition learned the material as well as, or better than, students in any other condition, but had the best understanding of the material, especially of causal and explanatory relationships.