Data on the survival of pollen ofTypha latifoliaL. stored for up to 261 d over seven different saturated salt solutions (providing 0.5 to 66% relative humidity) and six different constant temperatures (from −5 to +45 °C) were analysed to quantify the effect of air-dry storage environment on pollen longevity. Pollen survival curves conformed much more closely to negative cumulative normal distributions than to negative exponential relations. Estimates ofp50(storage period required to reduce pollen viability to 50%), provided by negative cumulative normal distributions, were available from 37 different storage environments in which pollen viability was reduced below 50%. Once observations at 0.5% and 5.5% relative humidity were excluded from analysis, there was a negative logarithmic relation between these estimates of longevity and pollen moisture content (%, wet basis) and a curvilinear semi-logarithmic relation between longevity and temperature. When the negative logarithmic relation between longevity and moisture content was replaced by a negative semi-logarithmic relation between longevity and the relative humidity of the storage environment the resultant model was less satisfactory, principally because pollen longevity over saturated solutions of calcium nitrate (43–62% relative humidity) and sodium nitrite (60–66% relative humidity) were consistently greater and smaller, respectively, than fitted values. Notwithstanding these errors, comparison between the fitted relations and observations at the two lowest relative humidities provided estimates of the lower-relative-humidity limits to these relations. These provisional estimates varied with storage temperature being lowest at 25 °C (<5.5% relative humidity). However, there was no linear trend to that variation (P>0.25): the mean estimate was 11.9 (s.e.=1.4)%. The considerable similarities among models of pollen longevity in air-dry storage, and their estimated lower limits, and those developed previously for orthodox seeds and spores are discussed.