The paper aims to apply bivariate probit model considering more than one explanatory variable for analyzing malnutrition data extracted from the 2004 Bangladesh Demographic and Health Survey (BDHS). Three estimation procedures namely maximum likelihood (ML), full information minimum chi-square (FIMC) and limited information minimum chi-square (LIMC) have been carried out for its parameters and also examined whether any difference existed among them in analyzing the data. An application of bootstrap method has also been provided to estimate the parameters of the model. In addition to this the paper also aimed to examine the effect of a few socio-demographic factors on malnutrition of children aged fewer than five in Bangladesh. Findings revealed that ML estimators are more efficient (with lower standard error) than two minimum chi-sqaure estimators: FIMC and LIMC in analyzing bivariate probit model for malnutrition data. Moreover. FIMC estimators are also found to be more efficient than that of LIMC. Furthermore, findings also revealed that breast feeding duration, birth order, sex of child, and area of residence have positive effect on children to be malnourished while household socioeconomic status and mother`s education have negative effect. Interpretation of findings has been explained with country`s socioeconomic point of view.