The present dissertation aims to contribute to the to date more limited findings of studies examining the factor of starting age in second language (L2) phonological learning in a formal instruction setting. Thus, its main objective was to assess the effects of starting ages of 8, 11, 14, and 18+ years on segmental perception and production of English as a foreign language (FL) in a strictly instruction-classroom context. Native Spanish and Catalan speakers learning English through schooling solely, who also differed in the amount of formal exposure to the FL (200, 416 and 726 hours of formal instruction in English), comprised the subject population under examination. All subjects were part of a long-term research project on the age factor in the acquisition of English as an FL in a formal instruction context - the Barcelona Age Factor (BAF) project - conducted at the Universitat de Barcelona over a period of 9 years (1995-2004). For the current study, subjects performed a same-different (AX) discrimination task and a production (word imitation) task.Results showed that in a formal learning environment starting age of 8 led to a better perception of English sounds in the long run than starting ages of 11, 14, and 18+. However, starting age effects on English segmental production both in the short- and in the long-term were inconclusive, as assessed by native English listeners. Similarly, exposure effects yielded mixed results as to more accurate perception and production of FL sounds. By contrast, a rather conclusive finding was obtained concerning subjects' first language (L1) dominance, namely being a Spanish or Catalan dominant speaker did not result in significantly more accurate English sound perception or production. Furthermore, female subjects were reported to produce English segments at more native-like levels than male subjects at all starting ages and with various degrees of formal exposure. Finally, it should be mentioned that the starting age effects observed in the short- and long-term (as conceived in the large research project design) provide invaluable evidence for characterising rate of acquisition of target language (TL) segmental perception and production by Catalan and Spanish native speakers (NSs) in a formal instruction learning context. The current dissertation concludes by suggesting a number of implications for further research.