The pediatric tumor neuroblastoma is a heterogeneous disease: Patients' clinical courses can range from spontaneous regression to fatal progression of the disease. Accordingly, treatment protocols vary from "wait and see" approaches to intensive multimodal therapies. Accurate risk estimation of the patients is therefore mandatory to choose the most adequate therapy. Current trials stratify by a limited number of clinical variables, such as stage of the disease and age of the patient at diagnosis, as well as molecular markers, such as amplification of the oncogene MYCN and loss of the short arm of chromosome 1. However, misclassifications of patients still occur, and thus, a precise prediction of the clinical courses remains a challenge of neuroblastoma research. In recent years, genomic alterations and gene expression profiles of this neoplasm have been characterized thoroughly. It has been shown that the diverse clinical phenotypes are reflected by both specific cytogenetic aberrations and distinct gene expression patterns. Moreover, a variety of DNA copy number changes and gene expression-based classifiers have been described that could predict the outcome of neuroblastoma patients more precisely than established prognostic variables. In this review, the recent advances in the detection and evaluation of molecular prognostic markers for neuroblastoma patients are summarized, and their current and potential contribution to risk stratification systems is discussed.