Measurements of the Sun's photospheric magnetic field can in principle be used to predict geomagnetic activity 1--3 days in advance. The accuracy of such predictions is low, however, because they do not include the north-south component of the interplanetary magnetic field (IMF Bz), which is the most geomagnetically relevant parameter. Aside from that carried by large-scale transients, IMF Bz is mainly a product of small-scale, in-transit turbulence and so is inherently unpredictable from solar measurements. Routine 1- to-3-day forecasts of geomagnetic activity based on deterministic algorithms are, therefore, not possible. Probabilistic forecasts offer the next best thing to deterministic forecasts, and air mass climatology offers a way to develop the advantages inherent in probabilistic forecasts for space weather applications. Here we address the IMF Bz indeterminacy problem (or better, get around it) by applying the concept of air mass climatology to the solar wind. We give criteria and statistics for solar wind air masses, which provide poof of concept for routine, midrange (1- to 3-day) probabilistic air mass forecasts of daily levels of geomagnetic activity.