Abstract: Direct spacecraft measurements of the solar wind and heliospheric magnetic field have been performed near-continuously for the over 60 years. The solar wind is seen to vary over all observed time scales, from subseconds to decades, with strong evidence of secular trends. In order to fully assess the longer-term variations, it is necessary to use proxy data. Approximately 170 years of geomagnetic data can be used to provide extremely accurate reconstructions of the near-Earth solar wind speed and the heliospheric magnetic field intensity on annual time scales, and consequently the open solar flux. Prior to 1845, it is necessary to go to more indirect proxies for solar wind conditions. Sunspot number can be used with simple empirical relations and open solar flux modelling to provide estimates of the near-Earth conditions back to ∼1610. Of course, such reconstructions are only as accurate as the sunspot records which underpin them, but by taking an ensemble approach, the uncertainties can be both reduced and quantified. Extending further back in time is possible through the use of cosmogenic isotope records in natural reservoirs such as tree trunks and ice cores. Teasing out the solar wind signal requires detailed modelling of the Earth system and hence the time resolution is generally sub-annual and the uncertainties greater than with geomagnetic and sunspot methods. Nevertheless, the long-term evolution of the solar magnetic field can be inferred over nearly 10,000 years.