The interpretation of mudrock Rb–Sr whole-rock regression ages using stratigraphically well-constrained samples from the Ordovician and Silurian of central Wales is re-examined. Two new Rb–Sr whole-rock regressions are presented along with K–Ar and Rb–Sr data from the <2 μm illite fraction of samples from these and published whole-rock suites. Eight K–Ar ages from the <2μm fractions give ages which show a normal distribution, with a mean and quadratically combined 2σ error indicating a Devonian age of 399 ± 3 Ma. Rb–Sr whole-rock ages are less consistent, giving 431 ± 10, 428 ± 14, 414 ± 6 and 403 ± 15 Ma. Rb–Sr data from the illite fractions do not generate a linear array. The Rb–Sr whole-rock isochron ages are interpreted as dating the time of passage of the rock through the smectite–illite transition because the radiometric ages correspond well to the estimated time of this reaction. Other explanations of the ages, such as the possibility that they were generated by mixing, or that they date sediment provenance ages, peak metamorphism, or cleavage, are rejected on geological grounds.