in this study reference as one of the five cohesive devices in the achievement of textuality in english and persian narrative/descripitive written texts is focused on and analysed . to do so , the theoretical framework elaborated by halliday and hasan (1976) and its version adapted by the writer to match the sub-types of reference in farsi are applied to the analysis of reference in english and persian texts. in the analysis of the texts, the distribution of reference sub-types are statistically verified and cross-compared with respect to within and between group comparison. therefore , the frequencies and percentages for the different types of reference items in the two languages are calculated . furthermore, the values of chi-sqare (x2) test are computed to determine whether or not the two languages differ from each other in their attempt to achieve texture through the use of different reference ties . statestically speaking, the results obtained from the analysis indicate that : 1) english and farsi are quite different regarding their employment of reference ties since the observed value of (x2) for the total number of reference items occurred in each language exceeds the critical value of (x2) . therefore, the null phypothesis (ho) that both languages display a similar pattern of reference utilization is safely rejected. 2) persian texts are much denser than english ones in terms of pronominals and demonstratives - the two of the three main types of reference. 3)both languages , however , use a similar amount of comparative items -the third main type of reference-to achieve texture. 4) english tends to employ pronominals and demonstratives almost equally; while, comparative items are , the lowest frequently used items among the reference ties in achieving texture. 5)like english, persian exhibits a general tendency to use an approximately equal number of pronominals and demonstratives . in farsi, too, these two kinds of reference sub-types play the most significant role in the realization of textuality. comparatives, on the other hand , take the least important position in reference distribution.