The concepts of cognate and false cognate are defined with reference to etymological considerations, reserving the term false friend for semantic problems. The importance of cognates as a language teaching/learning tool is considered, with special reference to the frequency of cognates in technical texts and their use as a vocabulary development strategy. Four problem areas are identified, namely: (a) low level of cognate recognition, (b) homonyms, (c) false cognates and false friends, and (d) the haphazard nature of cognate teaching. Analysis is made of a corpus of approximately 1500 English/Spanish cognates obtained from articles chosen at random from those journals most frequently consulted by students of computer science, engineering, and medicine, in order to ensure a representative sample of those cognates actually encountered by students in their daily reading activities. The cognates in the corpus are classified according to the following orthographical criteria: (a) individual letter equivalences, (b) position of differences, (c) vowel/consonant groups, and (d) part of speech. Results are given of tests administered to students at different levels to evaluate their ability to recognise and interpret cognates belonging to the classes established during the analysis of the corpus. It is suggested that teachers should adopt, as a regular part of class activities, a strategy of structured and graded practice in cognate recognition based on the analysis of the corpus and recognition tests.