This study examines the selected determinants of contraceptive prevalence among 10-49 aged ever-married women in Bangladesh and their true impact on the contraception prevalence rate (CPR). It applies a multilevel logistic regression analysis to draw valid conclusions about the effects of the selected determinants on CPR using the 2004 Bangladesh Demographic and Health Survey (BDHS) contraceptive binary data which is a multistage stratified cluster data. Instead of standard single level logistic model, multilevel logistic regression model has been utilized since the data follow a hierarchical structure. Also the comparison between single and multilevel model has been done to investigate the necessity of multilevel effects. The findings suggest that age of the women, number of living children, education, religion, media, place of residence and wealth index have significant multilevel effects on CPR. The study has finally suggested integrating a strong awareness program that targets the 10-49 aged currently married women in Bangladesh in those divisions (level-3) and clusters (level-2) where the particular determinant of contraceptive use has been found to be less effective.