We present a new approach to dating intermediate to mafic volcanic rocks using magnetite (U-Th)/He geochronology. Magnetite is common in volcanic rocks that typically do not contain easily datable minerals such as sanidine or zircon. Analytical procedures for producing magnetite (U-Th)/He ages have been developed, including mineral separation, sample air-abrasion to correct for a-ejection effects, He extraction/measurement, sample dissolution, and anion-exchange column chemistry procedures. Dated magnetite crystals were non-skeletal, euhedral to subhedral, and 100-300 mu m in size. To test the reliability of this new geochronometer, four basaltic to andesitic samples lacking sanidine or zircon were dated by both magnetite (U-Th)/He and whole-rock 40Ar/(39) Ar methods. For two samples, the ages from the different geochronometers are in excellent agreement (< 1 %). A third sample with a poorly behaved 40Ar/39Ar age spectrum affected by 39Ar recoil yielded a well-defined magnetite (U-Th)/He age that is consistent with 40Ar/(39) Ar age data from similar nearby volcanic rocks. The final sample, however, exhibited a near 40% discrepancy between the two methods, despite yielding reproducible magnetite (U-Th)/He ages. In all cases, the multi-aliquot magnetite (U-Th)/He ages (n > 7) exhibit 3-11% (2 sigma) variation about the mean age, indicating that reproducibility for magnetite (U-Th)/He ages is comparable to that of apatite and zircon (U-Th)/He analyses. In order to assess the He retentivity, we conducted a single magnetite helium diffusion experiment, yielding a well-behaved Arrhenius relationship and a closure temperature of similar to 250 degrees C (dT/dt= 10 degrees C/myr). Magnetite's high He retentivity coupled with (U-Th)/He age reproducibility demonstrates good potential for magnetite (U-Th)/He dating as an alternative volcanic geochronometer, particularly in cases where samples yield inconclusive or uninterpretable 40Ar/39Ar ages. (c) 2007 Elsevier B.V.. All rights reserved.