A long-standing belief is that statistical rules helpful in solving practical problems do not transfer beyond the subject matter domain in which they were learned. Recent research by G. T. Fong, D. H. Krantz, and R. E. Nisbett (1986) challenges this belief. Fong et al. showed that instructing learners about abstract rules, such as the law of large numbers, improved reasoning about ill-defined problems and transferred to solving everyday statistical problems that involved probabilistic relations. Fong et al.'s research is extended in 3 experiments with 276 university, secondary, and middle school students. The law-of-large-numbers heuristic was taught in regular classroom settings and students' abilities to solve ill-structured, everyday problems were tested. Students learned a good deal about how to reason statistically, and these gains generalized over different structures of problems and topics. The results support a revival of formalist views of transfer, that teaching formal rules about inference making can improve reasoning and support transfer.