Studies have indicated that location-based games can attain positive learning effects. However, the importance of the instructional design has been mostly overlooked. This paper aims at exploring the effects of discovery learning for location-based historic retrospection game. An experiment was conducted to analyze the learner's achievement and motivation. Discovery learning strategy and rote learning strategy were assigned to the experimental group and the control group respectively. Subjects were randomly assigned to one of the two in small groups of three. A pretest and a posttest were given to evaluate their learning achievement. A questionnaire was also given to investigate the learning motivation resulted from the two different instructional strategies. One month later, a delayed test was given to each group to explore the degree of learning retention. Initial data analyses indicated that different instructional strategies could result in different learning outcomes. We found that although the average learning achievement of discovery learning is not statistically different than that of rote learning, the resulted learning retention is statistically significant between the two groups. We conclude that learned information retains better through discovery learning strategy in a location-based game. This finding should prove valuable for instructional designers of location-based instructional game.