This thesis investigates the factors influencing the adaptation of foreign words to English, beyond traditional phonological parameters such as sonority distance. The data examined were produced in an on-line adaptation task to study purely linguistic rather than orthographic or historical influences. The adapted words contain only lesser-studied phonotactic problems rather than segmental ill-formedness. The choice of Russian as a donor and English as a borrowing language allow the study of adaptations in a setting which allows a further strategy of alteration of ill-formed consonant clusters beyond vowel epenthesis and consonant deletion, namely the substitution of segments to change one cluster into another. In contrast to previous research, English production of Russian stimuli with initial consonant clusters showed that segment change is applied frequently, comparable to the amount of vowel epenthesis. Extensive variation was observed, both in ratio of successful production, and in the choice and distribution of adaptation strategy. The factors in adaptation investigated were the sonority distance of the foreign clusters, as well as concepts which have received much recent attention within phonology, namely gradient grammaticality, similarity and frequency: English native speaker judgments were collected about the perceived grammaticality of foreign clusters and the similarity between targets and adaptations, while the frequency of possible adaptations in English was calculated from a corpus of spoken English. Results show that sonority cannot explain the variation in adaptation. Furthermore, frequency has no influence on the choice of adaptation; however, higher perceived badness results in a higher percentage of adaptations, and perceived similarity is decisive for the choice of adaptations. A comparison of similarity judgments of English and Russian listeners suggested that, in keeping with Steriade (2001), there are some cross-linguistically corresponding rankings of similarity; however, differences between languages due to phonotactics and phonetic detail were also found. In summary, the experiment results suggest that the adaptation of loanwords occurs in both in perception and production; furthermore, it is determined both by L1 specifics and cross-linguistic tendencies, an thus neither a straightforward application of L1 phonology nor completely independent of language background.