Within a 7-site prospective evaluation of the Bayer complexed prostate-specific antigen PSA (cPSA) assay, we analyzed the ability of cPSA to predict extracapsular extension (ECE) before radical prostatectomy. Included in this analysis were 152 men diagnosed with cancer, who subsequently underwent radical prostatectomy. Sera were tested with the Bayer total PSA (tPSA) and cPSA assays, and the Beckman free PSA (fPSA) and tPSA assays. Treating surgical pathology result as a binary variable (organ confined vs ECE), mean tPSA, cPSA, fPSA/tPSA (f/tPSA) ratios, tPSA density (tPSAD), and cPSA density (cPSAD) were compared by receiver operating characteristic (ROC) curves and univariate analysis. In all, 28 men (18.4%) had pathologically identified ECE. Between those with and without ECE, significant differences were observed for tPSA (P = 0.0127), cPSA (P = 0.0120), tPSAD (P = 0.0001), and cPSAD (P = 0.0002), but not f/tPSA (P = 0.3774) or c/tPSA (P = 0.2882). All tested parameters except f/tPSA (P = 0.376) and c/tPSA (P = 0.288) predicted ECE (P <0.05) by logistic regression. The ROC area under the curve (AUC) was identical for tPSA and cPSA (0.621) and for tPSAD (0.692) and cPSAD (0.691). Kendall-tau correlation coefficients also demonstrated the strongest correlation with ECE for cPSAD and tPSAD. Either alone or as a tPSAD calculation, cPSA carries equivalent staging ability to tPSA. The use of f/tPSA appears to be less effective in staging than either cPSA or tPSA, whereas the use of either cPSAD or tPSAD provides maximal staging accuracy. Therefore, cPSA could be applied as an accurate predictor of ECE independently or in a nomogram along with other predictive variables.