Microarray technologies have already provided valuable expression data in the classification of ovarian cancers based on gene profiling. However, the selection of patients for new therapeutic strategies remains a challenge. Low density arrays combine the capacity to measure the expression of many genes in a single sample, while retaining the sensitivity and quantitative range offered by qRT-PCR. An important limitation of high throughput techniques is the high quality requirements of starting RNA. The improvement of isolating RNA kits and the intrinsic characteristics of the qRTPCR approach has overcome the problem of using formalin-fixed, paraffin-embedded samples. The ability of real-time 28-demethyl-beta-amyrone RT-PCR to test the expression of very small mRNA fragments makes this assay affordable for studies with these kind of samples, in which the RNA is moderately or even highly degraded, and yet it still produces reliable results. Moreover, RT-PCR may be more feasible in the clinical setting than microarray-based technologies due to the need for specialized laboratory facilities and complex statistical analysis. We tried to obtain the maximum biological plausibility analyzing the expression of a group of genes involved in the same biological process by studying pathways implicated in the angiogenic process. To Imperatorin determine the gene expression patterns, RNA was extracted from 61 samples. We used Cox regression analysis based on the combination of significant genes for model selection. The Akaike Information Criterion was used to find the most accurate one. Rather than splitting data into test and validation sets, we performed a cross-validation, that uses repeated data-splitting to prevent model overfitting and to generate accurate estimates of model coefficients, being a compelling statistical technique for model validation. In the present study, the angiogenesis-related gene profile provided independent prognostic information for OS outcome in patients with advanced epithelial ovarian cancer. In addition, the profile allowed the differentiation of two groups with different PFS outcome.