Neuroblastomas is the most common extracranial solid tumour in children. The clinical behaviour of this tumour ranges from spontaneous regression to non-responsiveness to therapy. The molecular events that regulate the unique spontaneous regression, induced differentiation or heterogeneity in outcome within the same risk group are poorly understood. To develop a molecular basis for stratifying patients into various risk groups, we examined gene expression patterns that correlate with: (1) stage of the disease and (2) cellular differentiation of the tumour. We performed suppressive subtractive hybridisation (SSH) and microarray screening simultaneously, to compare and to combine the markers identified by both methods to develop a neuroblastoma specific array. The subtracted pool of 900 cDNA's were used to prepare a custom array. Unsubtracted and subtracted probes were used to screen neuroblastoma custom arrays and general microarrays representing known genes and EST's (10,000 cDNA's). A pool of 60 cDNAs (40% known and 60% novel genes) that were consistently differentially expressed were identified. The differential expression of 20 cDNAs was further evaluated using the Taqman sequence detection system based on real time RT-PCR quantitation. In preliminary studies, the expression of selected target genes correlated with the stage of disease and differentiation. The expression of known gene products such as cyclin A1 and p21 was further examined using immunohistochemistry, and correlated with mRNA levels. A large number of samples from retrospective clinical trials are currently being analysed for prognostic value of the identified targets using Taqman sequence detection and quantitation. Combining SSH and microarray screening will facilitate toward developing potentially disease specific and organ specific custom arrays that can be used for routine clinical stratification of patients.