The problem of variance estimation in two phase sampling under non response was not considered in the literature. We have adapted the classes of variance estimators suggested by Srivastava and Jhajj (1980) into the two phase sampling under the strategy given by Hansen and Hurwitz (1946) for the problem of non-response. In this paper some more generalize classes of estimators for the population variance are proposed using multi-auxiliary information in the presence of non-response in two phase sampling. Two situations of availability of auxiliary information and two situations of non-response are considered while constructing the classes. The expressions for mean square error are also provided and an empirical study is carried out to support the theoretical findings.