Multiple imputation framework, though originally developed for dealing with item missing values, can be used in many other problems when reframed as a missing data problem. Measurement error problem arises when the main study collects the data on the main outcome variables (Y) and mismeasured version of the main covariate of interest (X) and a sub-study where accurate (T) and missmeasured versions the covariate are available. Information from the sub-study can vary from simple descriptive statistics on the measurement error properties (T, X relationship) to a full fledged sub-study measuring all three aspects, (Y,T,X). Multiple imputation framework can be used to infer about the Y-T relationship by borrowing strength from both, the main study and sub-study. The standard multiple imputation combining rules, however, cannot be applied. New multiple imputation combining rules are developed and they differ across the types of sub-studies. The methodology is illustrated using actual data sets. A simulation study evaluates the repeated sampling properties of point estimates in terms of the bias and mean square error and coverage properties of the confidence intervals.