This author has mathematically correlated specific developmentpractices to defect density andprobability of on time delivev. Thispaper summarizes the results of this ongoing study that has evolved into a software prediction modeling and management technique. The author has collected datafrom 45 organizations developing software primarily for equipment or electronic systems. Of these 45 organizations, complete and unbiased delivered defect data and actual schedule delivery data was available for 17 organizations. The author willpresent in thispaper the mathematical correlation between the practices employed by these organizations and defect density. Thiscorrelation can and is used to: a) Predict defect density b) Improve software development practices for the best return on investment.