This novel approach allowed us to identify, for the first time in any vertebrate, a transcriptomic “fingerprint” evidencing specific molecular dysfunctions that is highly predictive of, and therefore likely to determine, egg quality. In the present study, we examined the relationship between reproductive dysfunction and the ovary transcriptome in domesticated striped bass, a species commonly exhibiting poor egg quality made manifest by early embryo developmental failure, which is a major problem in fish breeding programs. The female striped bass, which were matched for age, length and weight, were subjected to ovarian biopsy just before the breeding season and bred using standard hatchery procedures, and the quality of their spawns was judged by the percentage of eggs bearing viable embryos at 4 and 24 h following fertilization. All significant losses of embryos occurred in the first 4 h following fertilization, prior to MBT, during which time ontogeny is dependent on maternal gene transcripts stored within the ovulated egg. Remarkably, when mRNA extracted from the ovary biopsies was subjected to microarray and the data was analyzed using ANNs, models of the relation of ovary transcriptome to egg quality revealed that gene expression measured by as few as 250 probes, representing, 2% of the ovary transcriptome queried, could explain most variation in embryo survival. Furthermore, the ANN CV results showed that, 80% of variation in 4 h embryo survival could be predicted from gene expression measured using just the top 250 probe dataset. Not surprisingly, using additional probes did not improve model predictions, possibly because little scope in unexplained variation in egg quality remained, but it has been our experience that overly complex models train the ANN to the idiosyncrasies of the input data and are not robust to CV. Using fewer probes resulted in a decrease in the predictive ability of the models indicating that fertility depends upon the concerted action of many genes and it is quite possible to reduce the gene set so that model accuracy and robustness degrades. Scrutiny of gene expression patterns evaluated using the 233 most informative UniGenes revealed a conspicuous and unusual transcript profile associated with egg quality that was clearly evident on the fine resolution heat map of gene expression. In this profile, transcript abundance generally varies inversely between females producing high versus low quality spawns, although expression varies little for individual genes and, in the ANN models, individual transcripts make minor contributions to overall prediction of egg quality. Such small collective changes in magnitude of gene expression resulting in striking predictive effects have been similarly been reported for honeybee behavior. Interestingly, only one contig of the contigs identified as having particularly abundant ovary expression in our prior report.