Statistical matching methods goal is to combine several data sources to build datasets. The main goal of statistical matching is to make helpful and informative synthetic data without collecting more data or making new surveys. The study aims to use the goal programming approach in statistical matching to complete the data in two files, where the first file contains variables different from the second file, with one or more of the common variables. To reach this goal, a linear regression model is designed for each of the variables in each file in terms of the variables in the two files. The goal programming approach was used to estimate the parameters of the two regression models, and from it the estimated value of the variables presents in the first file and not present in the second file, and so on, hence we get a file with all the variables. The goal programming approach has the advantage of minimizing the effect of outliers with estimates because it uses minimization of the sum of absolute deviations. Moreover, the proposed approach has a constraint that guarantees significant estimations of the parameters. In addition to formulating the model, A simulation study evaluates the proposed approach's performance by generating and imputing data for dependent variables from different distributions. Results show the efficacy of the approach in accurately estimating missing values while maintaining data quality and minimizing errors.