To evaluate the erectile function (EF) domain of the International Index of Erectile Function (IIEF) as a diagnostic tool to discriminate between men with and without erectile dysfunction (ED) and to develop a clinically meaningful gradient of severity for ED. One thousand one hundred fifty-one men (1035 with and 116 without ED) who reported attempting sexual activity were evaluated using data from four clinical trials of sildenafil citrate (Viagra) and two control samples. The statistical program Classification and Regression Trees was used to determine optimal cutoff scores on the EF domain (range 6 to 30) to distinguish between men with and without ED and to determine levels of ED severity on the EF domain using the IIEF item on sexual intercourse satisfaction. For a 0.5 prevalence rate of ED, the optimal cutoff score was 25, with men scoring less than or equal to 25 classified as having ED and those scoring above 25 as not having ED (sensitivity 0.97, specificity 0.88). Sensitivity analyses revealed a robust statistical solution that was well supported with different assumed prevalence rates and several cross-validations. The severity of ED was classified into five categories: no ED (EF score 26 to 30), mild (EF score 22 to 25), mild to moderate (EF score 17 to 21), moderate (EF score 11 to 16), and severe (EF score 6 to 10). Substantial agreement was shown between these predicted and "true" classes (weighted kappa 0.80). The EF domain possesses favorable statistical properties as a diagnostic tool, not only in distinguishing between men with and without ED, but also in classifying levels of ED severity. Clinical validation with self-rated assessments of ED severity is warranted.