Theory of Mind (TOM) is a much researched area in developmental psychology. The evidence suggests that TOM development is universal and sequential, but both these claims require testing in different cultural groups, notably a lag in some small scale cultures (Vinden, 1999) and in Japanese children (Wellman, Cross, & Watson, 2001). Given the small number of non-Western data within the TOM corpus, present research was conducted with Pakistani preschoolers and comprised of three studies. The study 1 was carried out in two independent settings, one in school (Tryout 1) and other in home setting (Tryout 2). Tryout 1 included 72 preschoolers aged between 3-4 years to test, first, whether preschoolers in Pakistan showed the dominant pattern of failure in false belief at age 3 and success at age 4 (Wellman et al., 2001). In order to broaden the research framework beyond false belief, the children were also tested on Lillard and Flavell (1992) tasks of desire, pretence and belief. The results showed the expected significant age effect on false belief performance. However, 3 year olds were below statistical chance, while 4 year olds were at chance. In addition both age groups did not showed the expected patterns of performance in the Lillard and Flavell tasks. This significantly poor TOM performance was replicated in tryout 2 with 71 preschoolers (3-5 year olds) in home setting. Not only did 4 year olds lag behind their Western counterparts in all these tests, but the very different patterns of performance in this sample across a wide range of TOM skills underlines the need for more detailed analysis of the development of these skills in Pakistani preschoolers and possible education programmes to prepare children for the social demands of schooling. Despite the wide spread use of TOM tasks, determination of their psychometric properties remained neglected. Study 2 was aimed at calculating the vii test retest reliability of theory of mind (TOM) tasks (pretence, desire, belief, false belief) on 77 children from Islamabad. Two alternate forms of tasks were administered at two points in time, separated by approximately 2 weeks. Half of the children were administered with form 1 at time one and the other half were administered form 2 at time 1. Children performed at chance level on pretence, desire and belief tasks at both administrations. Performance was below chance for false belief (FB) tasks at both administrations. Moreover, Kappa values were satisfactory for pretence, desire and belief tasks (range = .43-.60), and the values fall below acceptable level for FB tasks. Study 3 was designed to investigate the concurrent and longitudinal associations of maternal and child’s language measures (content and quality) and children’s TOM (at time 2) over 8 months for 35 mother child dyads. Results indicated that maternal language quality (initiative talk at time 1) and child’s language content (cognitive terms at time 2) significantly predicted children’s TOM composite score (time 2). Moreover, maternal failed talk at time 2 significantly but negatively predicted: (a) belief task performance (b) other belief task performance (c) and composite TOM performance. However, child’s content of talk (cognitive terms at time 2) was an independent predictor of (a) composite TOM performance and (b) pretence task performance. The present study concurs with the need to eavesdrop on the familial language interaction to find out its contribution for the children’s TOM development. The implications of these findings for current research in language interaction and TOM are discussed.