Flood is a commonly occurring hazard in Malaysia. The climate change in combination with the sea level rise will affected the frequency of flood events especially in a tropical country like Malaysia. Many researches focused on modeling rainfall data have been carried out in Malaysia. However, most of the rainfall studies did not include the zero values. The importance of these zero measurements should be examined in order to increase the quality of the research. The main purpose of this paper is to study the effect of zero measurement in rainfall analysis by applying a mixed bivariate lognormal distribution. The inter-station correlation coefficient was calculated in three cases of datasets. The first case considered only the positive values at both stations, and the second case included the positive values at either one of the stations, while the third case considered all values including zeroes at both rainfall stations. It was found that only the cases considering the positive measurements are useful and valid for the characterization of rainfall fields in our analysis.