Abstract. Ice-nucleating particles (INPs) originating from deserts, semi-arid regions and dried lakebeds may cause heterogeneous ice nucleation impacting cloud properties. Recently, due to climate change and water scarcity, abandoned agricultural lands with little surficial crust and negligible vegetation cover have become an increasing source for atmospheric dust worldwide. Unlike deserts, these areas are rich in soluble salt and (bio-)organic compounds. Using soil samples from various sites of the Lake Urmia Playa (LUP) in northwestern Iran and airborne dusts collected at nearby meteorological stations we elucidate how minerals, soluble salts and organic matter interact to determine the IN activity of saline soils and dust. X-ray powder diffraction shows that the mineralogical composition is dominated by K-feldspars (microcline), quartz, carbonates, and clay minerals. The samples were stepwise stripped of organic matter, carbonates, and soluble salts. After each removal step, the ice nucleation (IN) activity was quantified in terms of onset freezing temperatures (Thet) and heterogeneously frozen fractions (Fhet) by emulsion freezing experiments using differential scanning calorimetry (DSC). We examined the influence of soluble salts and pH on microcline and quartz in emulsion freezing experiments, and comparing these with reference suspensions of microcline and quartz, exposed to salt concentrations and pH characteristic of the LUP samples. These analyses, combined with correlations between Thet and Fhet, allow us to identify the components that contribute to or inhibit IN activity. The LUP dusts turn out to be very good INPs with freezing onset temperatures around 248 K in immersion freezing experiments. Interestingly, their IN activity proves to be dominated by the relatively small share of (bio-)organic matter (1–5.3 %). These organic INPs are rich in carbohydrates, cellulosic and proteinaceous compounds, as determined by thermogravimetric analysis coupled with mass spectrometry (TGA-MS). After organic matter removal, the remaining IN activity (Thet ≈ 244 K) can be traced back to the clay fraction, because Thet and Fhet correlate positively with the clay mineral content but negatively with quartz and microcline. We attribute the inability of quartz and microcline to act as INPs to the basic pH of the LUP samples as well as to the presence of soluble salts, which lower ice nucleation temperatures below the freezing point depression due to the salt-induced reduction in water activity. This is confirmed by the reference freezing experiments with quartz and microcline suspensions. After additionally removing soluble salts and carbonates, the IN activity of the samples increases again significantly (Thet ≈ 249 K) and the negative correlation with quartz and microcline turned into a slightly positive one, which we explain by the recovered active sites of these minerals. Removing carbonates and salts from the natural samples leads to an increase in Thet and Fhet as well, indicating that their presence also suppresses the IN activity of the (bio-)organic INPs. Carbonates and soluble salts could do this, for instance, through occlusion of (bio-)organic INPs from the solution by a cementing effect, through their adsorption on the mineral surface, through interactions with dissolved Ca2+ undergoing complexation with the organic matter, or a combination of these effects. Overall, this study demonstrates that mineral and organic INPs do not just add up to yield the IN activity of soil dust, but that the freezing behavior is governed by inhibiting and promoting interactions between the components.