Objective: A refinement of prognostic variables using traditional pathologic markers integrated with oncogene proteins, enzymes, and hormonal factors may enhance the ability to predict for recurrence or survival in patients with mammary carcinoma. Although various oncogenes and oncogene products have been identified in human breast carcinoma, their relationship to disease outcome remains controversial. Methods: Using the monoclonal antibodies cS93.1, 9E1.0, F235-1.7.1, and PAb 1801 against each oncogene protein studied, the avidin-biotin complex immunoperoxidase method provided immunohistochemical staining of bound oncogene protein for c-fos, c-myc, Ha-ras, and p53, respectively. Analyses were made on archival pathology tissues of 85 breast cancer patients (stages I, IIA, and IIB). Forty patients (47%) had recurrence of disease; 45 remained free of local-regional or distant disease at mean follow-up of 48 months (range 6-180 months). Molecular biological data were merged with clinicopathologic demographics 1) to determine the frequency of single or co-expression of oncogenes in this patient population; 2) to evaluate the value of these molecular protein markers to predict probability of recurrence; and 3) to determine worth of the studied oncogenes to correlate with traditional clinical pathologic parameters and overall survival. Results: In this study, oncogene expression had statistical correlation for recurrence with increasing co-expression: one oncogene 17.2%, two oncogenes 56.3%, three or four oncogenes, 100% (p = 0.001). Increasing oncogene or co-oncogene expression correlated with statistically significant reduction in disease-free and overall survival; with no expression of oncogenes, disease-free survival was 30 (SE ± 5.7) months and overall survival was 56.4 (SE ± 4.57) months. With expression of three oncogenes, disease-free survival was 12 (SE ± 1.23) months (p = 0.0018) and overall survival was 23.4 (SE ± 3.38) months (p = 0.0025). In univariate Wilcoxon analysis, oncogene expression was the most significant variable to determine survival (p = 0.035); in multivariate analysis, age and oncogene co-expression each emerged as the most significant variables for overall survival. For the proportional hazards regression model, oncogene co-expression was significant (p = 0.0104, risk-ratio 1.914) and correlated with age and tumor size as significant variables. Ha-ras and c-fos both emerged as important individual oncogene proteins to affect survival (p = 0.0925, risk-ratio 3.517 and p = 0.025, risk-ratio 4.214, respectively). The proto-oncogene c-myc and the antitumor suppressor gene p53 did not have significant effects as individual oncogenes to influence survival. Conclusions: Approximately one fifth of the breast cancer patients in this analysis (disease-free and recurrent) expressed only a single oncogene marker (c-fos, c-myc, Ha-ras, or p53); one quarter of patients with recurrent disease expressed only one oncogene protein. Single oncogene expression did not possess independent prognostic significance for prediction of recurrence. Further, p53 mutations did not function as independent correlates for prognosis. The co-expression of the studied proto-oncogenes (c-myc, Ha-ras) and the nuclear transcriptional protein (c-fos) functioned as a strong prognostic correlate for recurrence and survival; the effect of individual oncogenes to predict survival was greatest for Ha-ras and c-fos. Immediate or early co-expression of three oncogene proteins in neoplastic transformation endowed cells of invasive carcinoma with an aggressive phenotype. This aggressive phenotype was evident in a small percentage of the studied population (11%) and predicted adverse disease-free and overall survival. These findings suggest that oncogene co-expression possesses significant prognostic and potential therapeutic value; incorporation of this molecular technology into future prospective randomized trials is advisable.