The basic features of the geomagnetic reversal chronology of the last 160 million years are well established. The relationship between this history and other features of the field, however, has been elusive. The determination of past field strength (paleointensity) is especially challenging. Commonly accepted results have come from analyses of bulk samples of lava. Historic lavas have been shown to faithfully record the past field strength when analyzed using the Thellier double-heating method. Data from older lavas, however, tend to show effects of in situ and laboratory-induced alteration. Here we review an alternative approach. Single plagioclase crystals can contain minute magnetic inclusions, 50--350 nm in size, that are potential high-fidelity field recorders. Thellier experiments using plagioclase feldspars from an historic lava on Hawaii provide a benchmark for the method. Rock magnetic data from older lavas indicate that the feldspars are less susceptible to experimental alteration than bulk samples. This resistance is likely related to the lack of clays. In addition, magnetic minerals are sheltered by the encasing silicate matrix from natural alteration that can otherwise transform the well-defined thermoremanent magnetization into an irresolute chemical remanent magnetization. If there is a relationship between geomagnetic reversal frequency and paleointensity, it should be best expressed during superchrons, intervals with few (or no) reversals. Thellier data sets based on single plagioclase crystals from lavas erupted during the Cretaceous Normal Polarity Superchron (~83--120 million years ago) suggest a strong (>12 ¿ 1022 Am2), stable field, consistent with an inverse relationship between reversal frequency and paleointensity. Superchrons may represent times when the pattern of core-mantle boundary heat flux allows the geodynamo to operate at peak efficiency, as suggested in some numerical models. Thellier data from single plagioclase crystals formed during times of moderate (10 reversals/million years) reversal occurrence suggest a weaker and more variable field. These paleointensity data, together with a consideration of paleomagnetic directions, suggest that geomagnetic reversals, field morphology, secular variation, and intensity are related. The linkages over tens of millions of years imply a lower mantle control on the geodynamo. On even longer timescales the magnetization held by plagioclase and other silicate crystals can be used to investigate the Proterozoic and Archean geomagnetic field during the onset of growth of the solid inner core. Data from plagioclase crystals separated from mafic dikes, together with directional data from whole rocks, indicate a dipole-dominated field similar to that of the modern, 2.5--2.7 billion years ago. Older Archean rocks are of great interestfor paleomagnetic and paleointensity investigations because they may record a time when the compositionally driven convection of the modern dynamo may not have been operating and a solid inner core did not play its current role in controlling the geometry of outer core flow. Most rocks of this age have been affected by low-grade metamorphism; investigations using single silicate grains provide arguably our best hope of seeing through secondary geologic events and reading the early history of the geodynamo. Absolute paleointensity measurements of the oldest rocks on the planet will require the further development of methods to investigate silicate crystals with exsolved magnetic minerals that address the uncertainties posed by thermocrystallization remanent magnetization, anisotropy, and slow cooling. Fortunately, prior work in rock magnetism, together with advances in analytical equipment and techniques, provides a solid foundation from which these frontier issues can be approached.