When the manufacturing sector plays a significant role in economic development, the demand for statistical data can only increase. Manufacturing data obtained from the annual Indonesian manufacturing survey from 1990 to 2010 cannot be used directly for analysis purposes due to data quality problems. This paper attempts to clean and balance the raw data by proposing several consecutive steps including the statistical modeling and coefficient of variant approaches. The results show that observations for 1,556 firms and 828 firms are obtained each year for the first period panel data and second period panel data, respectively; however, the results should be representative of the entire manufacturing sector in terms of the number of firms and sub-sectors. Furthermore, comparisons with existing data references are needed to verify the results.