We have previously shown that antibody−drug conjugates (ADCs) consisting of cAC10 (anti-CD30) linked to the antimitotic agent monomethylauristatin E (MMAE) lead to potent in vitro and in vivo activities against antigen positive tumor models. MMAF is a new antimitotic auristatin derivative with a charged C-terminal phenylalanine residue that attenuates its cytotoxic activity compared to its uncharged counterpart, MMAE, most likely due to impaired intracellular access. In vitro cytotoxicity studies indicated that mAb−maleimidocaproyl-valine-citrulline-p-aminobenzyloxycarbonyl−MMAF (mAb−L1−MMAF) conjugates were >2200-fold more potent than free MMAF on a large panel of CD30 positive hematologic cell lines. As with cAC10−L1−MMAE, the corresponding MMAF ADC induced cures and regressions of established xenograft tumors at well tolerated doses. To further optimize the ADC, several new linkers were generated in which various components within the L1 linker were either altered or deleted. One of the most promising linkers contained a noncleavable maleimidocaproyl (L4) spacer between the drug and the mAb. cAC10−L4−MMAF was approximately as potent in vitro as cAC10−L1−MMAF against a large panel of cell lines and was equally potent in vivo. Importantly, cAC10−L4−MMAF was tolerated at >3 times the MTD of cAC10−L1−MMAF. LCMS studies indicated that drug released from cAC10−L4−MMAF was the cysteine−L4−MMAF adduct, which likely arises from mAb degradation within the lysosomes of target cells. This new linker technology appears to be ideally suited for drugs that are both relatively cell-impermeable and tolerant of substitution with amino acids. Thus, alterations of the linker have pronounced impacts on toxicity and lead to new ADCs with greatly improved therapeutic indices.