Aim: the aim of this study is determine the sensitivity Digit Span, Word Span, Non-Word Span tests on Assessment of working memory in children and the psychometric properties of these tests. Method: Ex post facto research method and the population of all elementary school students of Tehran district one and the number was 83345 people. Among them through purposeful sampling, 178 male and female students studying In the second semester 2014-15 at elementary schools of Tehran district one (Including schools: Farzandan Iran zamin, Farzane 2, Edalat 1, Zahha) With regard to the criteria for entry and exit Research Including lack of psychiatric drugs, lack of physical illness and informed consent to participate in the study were selected. The Research tools include tests: Digit Span (Gathercole & Pickering, 2000), word (Marconi, Ott, Pesenti, Ratti & Tavella, 1993) and nonword (Gathercole, Willis, Baddeley & Emsile, 1994). Data using discriminant analysis and analysis of multivariate variance were analyzed. Results: The results showed that non-word span test had the highest correlation with diagnostic function and is more reliable in recognizing the dextrally (F = 14. 31, P<0. 009). Also, non-word span test has a suitable psychometric sensitivity in measuring working memory in children along with age increase (F = 22. 36, P<0. 006). Conclusion: According to the results of this study, non-word span test included better psychometric features comparing with span word and digit span tests. This test could be used in research and clinical settings.