Cancer is the second leading cause of death among children, second only to accidents. For children, the most commonly occurring cancer is leukemia. A prominent phenotypic abnormality of human acute leukemia is the inability of the cells to differentiate into functional mature cells; instead, they are blocked at an early stage of development and remain in the proliferative pool and rapidly accumulate. Today common approaches for the cure of acute leukemia consist in an aggressive chemotherapy treatment with a strong “cocktail” of drugs. These treatments are effective but beget several side effects and toxicity. Therefore, it is important to design new efficient drugs with fewer side effects and acting through other mechanisms. Preliminary studies have identified as lead compound a new antileukemic agent consisting of a 5α-androstane-3α,17β-diol as steroid skeleton and a N-methylpiperazine nucleus at position 2β. To speed up the development of this new family of potential drugs, we generated libraries of derivatives using solid-phase organic synthesis (SPOS). Starting with dihydrotestosterone, a 5-step sequence of reactions afforded the 2,3-alkene steroid. After treatment of this alkene with m-chloroperbenzoic acid, the resulting 2α,3α-epoxide was selectively opened with piperazine to give, after N-Fmoc protection, an aminosteroid with the suitable 2β-piperazino and 3α-hydroxy stereochemistry as demonstrated by NMR spectroscopy and X-ray analysis. The difference of reactivity between steroidal 3α and 17β OHs was then used to allow the regioselective 17β-OH coupling of steroid to polystyrene-butyldiethylsilane (PS-DES) resin in 80% yield. The presence of a steroid linked to PS-DES resin was confirmed by characteristic IR and NMR signals. From this key precursor, the solid-phase synthesis began with the cleavage of the Fmoc protecting group and the addition of amino acids and carboxylic acids by acylation to introduce two or three levels of molecular diversity. An acid cleavage allowed generating three libraries of 25, 25 and 80 individual compounds with 1, 2 and 3 levels of molecular diversity, respectively. The three libraries were submitted to a random sampling and the selected members characterized by TLC, 1H NMR and MS analysis. These results confirmed the reactivity of all building blocks used in the elaboration of each library and allowed us to validate the rest of the library members. The overall mean crude yields for the solid-phase sequence of reactions were 75% for library A (6 steps), 94% for library B (6 steps) and 93% for library C (8 steps). The purity of aminosteroids released from the resin was estimated by TLC and NMR and found to be 70-90% for library A, 80-90% for library B and 50-80% for library C. In order to identity some hits, all members of libraries A-C were then submitted to a preliminary antiproliferative assay in human myeloid leukemia HL-60 cells. Several members of these libraries were more potent agents than the lead compound, especially three members having the amino acid proline as the first level of diversity and a cyclohexylcarbonyl (B10), a methylbutyryl (A7) or a cyclohexylacetyl (A9) as the second level of diversity. After we identified the biological potential of aminosteroids A7, A9 and B10, they were synthesized using the same approach discussed above, but in larger amounts for purposes of purification, characterization and validation of biological activity. The three compounds have a proline as amino acid element of diversity, suggesting its important contribution to the cytotoxic activity. The yields (∼65% for 5 steps after final purification) and purity obtained after this second synthesis were representative of the previously synthesized compounds. Representative aminosteroids A7, A9 and B10 were fully characterized by IR, NMR and MS analysis, and 2D NMR experiments (APT, HSQC, HMBC, COSY and NOESY) confirmed their structure. With pure compounds in our hands, we determined their IC50 values against HL-60 cells. Compounds A7 and B10 showed a better antiproliferative activity than A9 since they displayed an IC50 of 0.66 ± 0.07 and 0.58 ± 0.07 μM, respectively, compared to 1.78 ± 0.21 μM for A9. Furthermore, they are selective for cancer cells since they are non-toxic toward normal human cell line WI-38. This selectivity is important since standard chemotherapeutic agents, such as doxorubicin, are toxic on normal cells and cause severe side effects.