The present study is an investigation of communication strategies (CSs) by EFL learners at Sultan Qaboos University (SQU). More specifically, it addressed three issues which can be stated as follows: (1) the type and frequency of CSs as determined by the descriptive and t test techniques. (2) the influence of type of task (namely story telling and translation) on the use of CSs. (3) the impact of the learners' L2 proficiency (instructional) level on the use of CSs. The population for our study was that of EFL learners at SQU and the sample consisted of 90 students from the English program in the college of Education and Islamic Sciences in this university. The subjects were selected from three instructional levels: second year (semester 3), third year (semester 5) and fourth year (semester 7). These levels constituted one of the independent variables of the study. The materials for data collection consisted of two tasks, namely story telling and translation, both of which dealt with the same topic and targeted a set of pre-determined key concepts. In these tasks, the subjects were asked to produce prose materials, following task-specific instructions. The subjects' production was then analyzed with the purpose of identifying the CSs used and their frequency of occurrence. This analysis was guided by a pre-determined taxonomy of CSs selected from the literature. The reliability of the classification of CSs was established through the inter-rater technique in which a sample of CSs was analyzed by two judges (in addition to the researcher). The three classifications were then subjected to the correlation technique. The value of the correlation ranged from .81 to .88. The findings of the study can be summarized as follows: 1) The strategies investigated were used by the subjects to varying degrees. Among other things, the achievement strategy was used significantly more often than the reduction strategy when the two tasks were considered together. The most commonly used strategies were: approximation, semantic avoidance, superordination, generalization and paraphrase. The least commonly used ones were: word coinage, circumlocution, literal translation and message abandonment. 2) The nature of the task significantly affected the learners' choice of CSs. For example, approximation, generalization, paraphrase and message abandonment were used significantly more often in story telling than in translation. 3) The choice of CSS was significantly affected by the learners' proficiency (instructional) level; however, no real developmental pattern was shown in this choice across the levels. The implications suggested by the study included: (a) a focus in future research on the CSs that characterise the interlanguage of Arabic speaking learners of English, (b) the need for a more extensive. investigation of the variations in CSs that can result from other types of writing tasks and a variety of types of content. With regard to the pedagogical implications, it was recommended that more attention could be given in EFL instructional materials to the teaching of CSs.