Two series of nonclassical antifolates (2,4-diamino-5-deaza compounds 2-5 and 5,10-dideaza compounds 6-13) were synthesized as inhibitors of dihydrofolate reductase (DHFR) from Pneumocystis carinii (pc) and Toxoplasma gondii (tg) organisms that are responsible for fatal opportunistic infections in AIDS patients. Rat liver (rl) DHFR served as the mammalian reference enzyme to determine selectivity. Syntheses of the target 5-deaza compounds were achieved by initial construction of the pivaloyl-protected 2,4-diamino-6-bromopyrido[2,3-d]-pyrimidine 17 via a cyclocondensation of 2,4,6-triaminopyrimidine with bromomalonaldehyde. Sequential Heck coupling of 17 with styrene followed by ozonolysis afforded the 6-formyl derivative 19. Reductive amination of 19 with 3,4,5-trimethoxyaniline afforded the N10-H analog. The N10-Me and N10-Et analogs were synthesized by nucleophilic displacement of the 6-bromomethyl derivative 22 (obtained from the 6-formyl derivative 19 by reduction and bromination) with the appropriate N-alkylaniline. The trans-5,10-dideaza analogs 6-8 were synthesized via a Heck coupling of the appropriate methoxystyrene with 17, and selective reduction of the resulting 9,10-double bond afforded target compounds 9-11. Further reduction to the tetrahydro derivatives afforded analogs 12 and 13. The 5-deaza N10-Me 3,4,5-trimethoxy analog 3 maintained the best balance of potency and selectivity against both tgDHFR and pcDHFR. Compared to trimethoprim, compound 3 was only slightly less selective but was 300-fold more potent against tgDHFR. The 5,10-dideaza analogs were generally less potent and selective than the 5-deaza compounds.