This study examines two ways of achieving long-term vocabulary retention: learning words through context versus traditional bilingual (L1-L2) paired associates. It compares the results of the two vocabulary teaching methods received by a research in a secondary school of economics. The participants consisted of 40 native speakers of Hungarian learning English as a foreign language, aged 15-16, students attending grade 9. All the subject share similar education background, they all belong to beginners’ groups following some years of English language learning in the primary school. During the research, Group1 of 21 participants learned meaning of words inferred from context and definitions, while Group2 of 19 participants applied L2-L1 word lists. The procedure started with a pre-test to investigate the students’ word knowledge. Then in the teaching phase, the presentation and the elaboration was directed by the two different methods. The participants’ vocabulary achievement was measured by an immediate post-test with vocabulary recognition tasks. Two weeks later a delayed post-test containing both production and recognition tasks was administered to get results about the long-term vocabulary retention. Three hypotheses of the study were formulated: two of them on the immediate post-test results, namely that each group will attain significantly higher scores in the task, which is closer to its learning conditions. The main hypothesis of the study was connected to the final goal of vocabulary teaching and learning: long-term vocabulary retention. The researcher of the study presupposed that Group1 (with context-embedded word-meaning input) would outperform Group2 (with the input of a bilingual wordlist) on the delayed test measuring long-term vocabulary retention. The hypothesis was based on the supposition that vocabulary learning by inferred meaning gives better results than by bilingual paired associates in long term due to the greater level of depth of processing, elaboration and mental effort. On the receptive sentence completion test measuring immediate vocabulary acquisition, students with meaning-enhanced input significantly outperformed those with input through wordlists. On the receptive L2-L1 meaning recognition test, students with wordlist input did not outperform significantly the other group; thus, the hypothesis is disproved. The main hypothesis concerning long-term retention is proved by the result of the analysis. It confirms that students learning English words meaning-enhanced and context-embedded from definition in the target language can achieve higher level of long-term vocabulary retention than those learners who are given bilingual wordlists.