QUANTITATIVE ASSESSMENTS OF SOFTWARE RELIABILITY.

After stressing the current need for quantitative measures of software reliability for prediction, monitoring, and a posteriori assessment purposes, the paper provides a categorization and a description of those approaches leading to practical applications. Three main approaches to quantitatively assessing software reliability are identified, namely, models adapted from classical reliability theory wherein reliability is expressed in terms of software error rate; models based on sampling techniques applied to the error domain or to input data domain and models based on program complexity measures. Underlying assumptions and areas of application are indicated.