We have studied the molecular basis of drug resistance in human CCRF‐CEM leukemia cells exposed to high dose intermittent pulses of novel polyglutamatable antifolates that target various folate‐dependent enzymes. These include the dihydrofolate reductase (DHFR) inhibitors edatrexate, methotrexate and aminopterin, the thymidylate synthase (TS) inhibitors ZD1694 and GW1843, the glycinamide ribonucleotide formyltransferase (GARTF) inhibitor DDATHF as well as the multitargeted antifolate LY231514 inhibiting both TS, DHFR and GARTF. Fourteen antifolate‐resistant sublines were isolated, 11 of which displayed a drug resistance phenotype that was based on impaired folylpoly‐γ‐glutamate synthetase (FPGS) activity as these cell lines: 1) typically lost 90–99% of parental FPGS activity; 2) expressed 1.4–3.3‐fold less FPGS mRNA (only 4 cell lines); 3) displayed up to 10⁵‐fold resistance to polyglutamylation‐dependent antifolates including ZD1694 and MTA; 4) retained sensitivity to polyglutamylation‐independent antifolates including ZD9331 and PT523; 5) were up to 19‐fold hypersensitive to the lipid‐soluble antifolates trimetrexate and AG377; 6) had a normal or a small decrease in [³H]MTX transport; and 7) had a 2.1–8.3‐fold decreased cellular folate pools and a consequently increased folate growth requirement. The remaining 3 antifolate‐resistant sublines lost 94–97% of parental [³H]MTX transport and thus displayed a high level resistance to all hydrophilic antifolates. To screen for mutations in the hFPGS gene, we devised an RT‐PCR single strand conformational polymorphism (SSCP) assay. RT‐PCR‐SSCP analysis and DNA sequencing showed that only a single FPGS‐deficient subline harbored an FPGS mutation (Cys346Phe). Three‐dimensional modeling of the human FPGS based on the crystal structure of Lactobacillus casei FPGS suggested that this mutation maps to the active site and interferes with the catalytic activity of the enzyme due to a putative bulky clash between the mutant Phe346 and a native Phe350 within α‐helix A10 in a highly conserved C‐terminal hydrophobic core. This was consistent with a 23‐fold decreased affinity of the mutant Cys346Phe FPGS for L‐glutamate. We conclude that decreased FPGS activity is a dominant mechanism of resistance to polyglutamylation‐dependent novel antifolates upon a high‐dose intermittent exposure schedule. The finding that cells may exhibit 5 orders of magnitude of resistance to polyglutamylation‐dependent antifolates but in the same time retain parental sensitivity or hypersensitivity to polyglutamylation‐independent antifolates or lipophilic antifolates offers a potentially promising treatment strategy in the overcoming of FPGS‐based anticancer drug resistance. © 2002 Wiley‐Liss, Inc.