A major unresolved question in keyword mnemonics is whether the technique is more effective when the keyword is supplied by the researcher, or when it is generated by the participants. We have recently described a hybrid approach: keywords are selected by the researcher from among those generated by persons of similar sociodemographic characteristics to those of the persons with which the technique is to be used. In the present study we investigated whether recall is affected by the degree of among-persons concordance at the keyword generation stage. The 109 participants (12-to 15-year-olds) received a list of 10 Latin words, each with keyword (previously generated by persons of similar sociodemographic characteristics), translation to Spanish. For five of the 10 Latin words, there had been high among-subject concordance at the keyword generation stage, whereas for the remaining five concordance had been low. Our results indicate that both immediate and one-week recall were significantly better for words for which there had been high concordance in keyword selection. These results indicate ways in which the keyword method can be used more effectively to aid word learning.The term "key word mnemonics" was coined by Atkinson [1] in reference to a mnemonic rule successfully used for Russian vocabulary learning by Englishspeaking students. In application, the technique can be divided into two stages, the first verbal and the second visual. In the first stage, the aim is to generate a familiar, 347 Ó 2005, Baywood Publishing Co., Inc.