In the 90 days following the Oroville, California, earthquake (August 1, 1975; ML = 5.7), 313 positively identified strong-motion accelerograms were obtained for 86 different aftershocks. This set of records samples a wide range of magnitude (1.8 ≦ ML ≦ 5.2), focal depth (2 ≦ h ≦ 12 km), and site conditions (Holocene alluvium to Mesozoic crystalline basement). Most of these records were written at hypocentral distances R ≦ 15 km, a distance range for which relatively few strong-motion accelerograms had previously been available. Equally significant is the completeness of coverage (seven or more positively identified records) available for each of 12 well-recorded aftershocks (2.8 ≦ ML ≦ 4.9). One hundred and sixty-five peak acceleration values at 6.7 ≦ R ≦ 15 km for 33 aftershocks (3.0 ≦ ML ≦ 4.9) are the basis of an investigation of the dependence of peak acceleration on magnitude at R ≃ 10 km. These observations are supplemented with one datum for the main shock, 6 observations for two ML ≧ 5 aftershocks, and 61 null observations. Peak accelerations at sites on bedrock or minor sedimentary thicknesses are consistently higher than those at sites on several hundred meters or more of sedimentary rocks. Mean values for both site classes of data increase fairly rapidly with magnitude for 3 ≦ ML < 5, but the little available data for ML ≧ 5 suggest this magnitude dependence abruptly terminates at ML ≃ 5. The magnitude dependence of the sedimentarysite data is apparently stronger than for the bedrock-site data. Much if not all of the apparent magnitude dependence of these data can be attributed to properties of the path and not the source, consistent with the hypothesis that the amplitude of the peak acceleration phase, as it leaves the source region, does not depend on source strength although the predominant frequency of this phase increases as the duration and magnitude of the earthquake decreases.