Two experiments evaluated pictorial symbols intended to warn consumers of a child chokinghazard. In Experiment 1, participants used the Zwaga (1989) comprehension estimation procedure to estimate the percentage of the U.S. population they felt would understand the intended meaning of IS symbols designed to depict a choking hazard. For the symbol with the highest comprehension estimate, participants indicated that 74% of the population would correctly interpret the intended meaning: choking. All of the top-rated symbols contained similar combinations of features, including crossedhands, a protruding tongue, and a frontal face orientation. But interestingly, symbols with lower comprehension estimates also contained a different set of similar features. In Experiment 2, small groups of participants constructed multiple panel choking hazard symbols from the set of 15 symbols in the first experiment. As in the first experiment, there was a high degree of consistency among participants' responses. Unexpectedly, the first and last symbols selected for this three-panel sequence were among symbols that received the lowest comprehension estimates in the first experiment. These fmdings seem to confirm that symbol comprehensibility can be influenced by context. Implications ofthis study for future research are discussed.