In this study, 9th-grade students (N = 196) with a mean age of 14.7 years read a scientific text explaining the chemical process of doing laundry with soap and water and then took 3 tests. Students who were instructed to generate drawings during learning scored higher than students who only read on subsequent tests of transfer (d = 0.91), retention (d = 0.87), and drawing (d = 2.00). For students who were instructed to generate drawings during learning, those who generated high-accuracy drawings (according to a median split) scored higher than students who generated low-accuracy drawings on subsequent tests of transfer (d = 0.99), retention (d = 0.79), and drawing ( d = 1.87); furthermore, drawing-accuracy scores during learning correlated with learning-outcome scores on transfer ( r = .57), retention ( r = .50), and drawing ( r = .82). Results suggest that drawing can serve as a generative activity and as a prognostic activity.