Abstract We report 39Ar–40Ar ages of whole rock (WR) and plagioclase and pyroxene mineral separates of nakhlites MIL 03346 and Y-000593, and of WR samples of nakhlites NWA 998 and Nakhla. All age spectra are complex and indicate variable degrees of 39Ar recoil and variable amounts of trapped 40Ar in the samples. Thus, we examine possible Ar–Ar ages in several ways. From consideration of both limited plateau ages and isochron ages, we prefer Ar–Ar ages of NWA 998 = 1334 ± 11 Ma, MIL 03346 = 1368 ± 83 Ma (mesostasis) and 1334 ± 54 Ma (pyroxene), Y-000593 = 1367 ± 7 Ma, and Nakhla = 1357 ± 11 Ma, (2σ errors). For NWA 998 and MIL 03346 the Ar–Ar ages are within uncertainties of preliminary Rb–Sr isochron ages reported in the literature. These Ar–Ar ages for Y-000593 and Nakhla are several Ma older than Sm–Nd ages reported in the literature. We conclude that the major factor in producing Ar–Ar ages slightly too old is the presence of small amounts of trapped martian or terrestrial 40Ar on weathered grain surfaces that was degassed along with the first several percent of 39Ar. A total K–40Ar isochron for WR and mineral data from five nakhlites analyzed by us, plus Lafayette data in the literature, gives an isochron age of 1325 ± 18 Ma (2σ). We emphasize the precision of this isochron over the value of the isochron age. Our Ar–Ar data are consistent with a common formation age for nakhlites. The cosmic-ray exposure (CRE) age for NWA 998 of ∼12 Ma is also similar to CRE ages for other nakhlites.