AKR1C3 is an enzyme that is overexpressed in several types of radiotherapy- and chemotherapy-resistant cancers. Despite AKR1C3 is a validated target for drug development, no inhibitor has been approved for clinical use. In this manuscript, we describe our study of a new series of potent AKR1C3-targeting 3-hydroxybenzoisoxazole based inhibitors that display high selectivity over the AKR1C2 isoform and low micromolar activity in inhibiting 22Rv1 prostate cancer cell proliferation. In silico studies suggested proper substituents to increase compound potency and provided with a mechanistic explanation that could clarify their different activity, later confirmed by X-ray crystallography. Both the in-silico studies and the crystallographic data highlight the importance of 90° rotation around the single bond of the biphenyl group, in ensuring that the inhibitor can adopt the optimal binding mode within the active pocket. The p-biphenyls that bear the meta-methoxy, and the ortho- and meta-trifluoromethyl substituents (in compounds 6a, 6e and 6f respectively) proved to be the best contributors to cellular potency as they provided the best IC50 values in series (2.3, 2.0 and 2.4 μM respectively) and showed no toxicity towards human MRC-5 cells. Co-treatment with scalar dilutions of either compound 6 or 6e and the clinically used drug abiraterone led to a significant reduction in cell proliferation, and thus confirmed that treatment with both CYP171A1-and AKR1C3-targeting compounds possess the potential to intervene in key steps in the steroidogenic pathway. Taken together, the novel compounds display desirable biochemical potency and cellular target inhibition as well as good in-vitro ADME properties, which highlight their potential for further preclinical studies.