Ninety-nine red bed samples were collected from 15 sites of the Rose Hill Formation in the central Appalachians of West Virginia and Virginia. These hematite-bearing samples were demagnetized with thermal and chemical techniques which gave mutually consistent results. Most samples had reversed polarity directions which in some cases are undoubtedly of secondary (late Paleozoic) origin. Other reversed directions may be of Middle Silurian age, but this cannot be unambiguously demonstrated, since corrections for the tilt of the strata do not always provide a conclusive fold test. On the other hand, 17 samples showed normal polarity directions with northerly declinations. Some of these samples contained only one stable component of magnetization, whereas others reached a stable endpoint only after dominant reversed polarity components of magnetization had been removed by thermal demagnetization. Five of these samples did not reach a stable endpoint. The northerly directed components of normal polarity show a marked improvement in grouping after unfolding, indicating that these directions were acquired before the Appalachian folding. Their mean direction of magnetization, with declination 344¿ and inclination -42¿ (k=10.4, α95=14¿, and N=12), yields a pole at 26¿N, 116¿E, which is close to published poles for North America of Silurian and Ordovician age. Although this pole should not be considered as firmly established, we consider it to be more reliable than pole positions obtained (through whatever selection criteria) from the reversed polarity samples. Apparently, the directions and blocking temperature spectra of the secondary and reversed polarity original magnetizations are too similar to allow their separation or recognition in thermal demagnetizations.