Techniques are described for determining the quality of software using fuzzy logic. The techniques include using linguistic variables to translate a numeric value of a quality attribute into a linguistic value of a quality attribute. The linguistic values of multiple quality attributes are aggregated into a quality rating of the software. The quality rating then is used to predict a performance attribute for the software. Techniques are described that may help improve the accuracy of the prediction, including refining software quality parameters based on a comparison of the parameters with software quality of the developed software. The techniques also include the ability to modify the quality attributes used based on the computing environment, the organizational environment or other characteristics.