Fusarium oxysporum is one of the most common species causing soybean root rot and seedling blight in the U.S. In a recent study, significant variation in aggressiveness was observed among isolates of F. oxysporum collected from roots in Iowa, ranging from highly pathogenic to weakly or non-pathogenic isolates. In the present work, a RNA-seq-based analysis was used for the first time to investigate the molecular aspect of the interaction of a partially resistant soybean genotype with non-pathogenic/pathogenic isolates of F. oxysporum at 72 and 96 hours post inoculation (hpi). Markedly different gene expression profiles were observed in compatible and incompatible host-pathogen combinations. A peak of differentially expressed genes (DEGs) was observed at 72 hpi in soybean roots in response to both isolates, although the number of DEGs was about eight times higher for the pathogenic isolate compared to the non-pathogenic one (1,659 vs. 203 DEGs, respectively). Furthermore, not only the number of genes, but also the magnitude of induction was much greater in response to the pathogenic isolate. This response included a stronger activation of many well-known defense-related genes, and several genes involved in ethylene biosynthesis and signalling, transcription factors, secondary and sugar metabolism. In addition, 1130 fungal genes were differentially expressed between the F. oxysporum isolates in planta during the infection process. Interestingly, 10% of these genes encode plant cell-wall degrading enzymes, reactive oxygen species-related enzymes and fungal proteins involved in primary metabolic pathways. Such information may be useful in the development of new methods of broadening resistance of soybean to F. oxysporum, including the silencing of important fungal genes, and also to understand the molecular basis of soybean-F. oxysporum interactions. Overall design: Soybean seedlings mRNA profiles inoculated with a non-pathogenic and pathogenic isolates of F. oxysporum and collected at 72 and 96 hpi, were generated using Illumina HiSeq 2500. Control seedlings were also included for each time of inoculation. Three biological replicates were considered for each condition, 18 samples in total.