Two experiments compared the effectiveness of two separate mnemonic devices for learning the states and their capitals--one a complex key word system using substitute words for each syllable, the other a simple key word device interacting key words from the state and capital names in an illustration. In the first experiment, 88 fourth and fifth grade students were randomly assigned to 1 of 4 experimental conditions to learn the capitals of 14 states. Students in the control group used their own "best method" for learning the capitals, those in the second group used the complex key word device, and those in the third and fourth groups used a one-stage (similar to the complex device) and a three-stage version of the simple key word device. Students were tested for immediate recall after the session and again the following day for delayed recall. In the second experiment, 59 fourth grade students were assigned to control, simple key word, or complex key word groups. The procedures paralleled those in experiment 1 except that in both experimental groups the capital key word learning stage contained an additional trial to raise the level of subsequent recall. Students were tested immediately and 2 days later. The results of both experiments indicated that the purportedly effective complex key word method was no more effective than students' own devices when subject to controlled experimentation. On both the immediate and delayed recall measure, simple key word students surpassed both control and complex key word device students. (HTH)