Gene expression during early stages of fiber development is poorly understood. Here we report the development of a full-length cDNA library derived from G. hirsutum L. Texas Marker-1 (TM1) immature ovules (TMO) from -3 to 3 days pre/post anthesis (DPA). The library contained a total of 4x106 colonies including >60% full-length cDNAs. We generated 32,789 ESTs with an average length of 763-bp, of which ~78% ESTs had a sequence longer than 700 bp. The TMO ESTs were assembled into 8,272 unique sequences including 4,036 tentative consensus (TC) sequences and 4,504 singletons, representing ~20% unique sequences in the cotton EST collection. Compared to ~128,000 existing ESTs derived from elongated fibers and non-fiber tissues, TMO ESTs showed significant increase in the percentage of genes encoding putative transcription factors and other regulators involved in trichome development such as WRKY, AP2/EREBP, C2H2, MYB, and bHLH transcription factors and proteins involved in auxin, brassinosteroid, gibberellic acid, and ABA signaling pathways. These activities coincide with the predicted biological activities during fiber cell initiation. With the addition of the new ESTs, the Cotton Gene Index was rebuilt using a total of ~160,000 ESTs that contained genome-specific SNPs and SSRs, many of which may be developed as portable molecular markers for cotton breeding. Furthermore, in silico analysis of gene expression identified a large number of genome-specific genes that were differentially enriched in 5 different non-normalized cotton cDNA libraries derived from fiber and non-fiber tissues. The expression patterns of selected ovule-specific ESTs were confirmed by qRT-PCR analysis. The data reveal important molecular processes concomitant with the early stages of cotton fiber development.