Iron (Fe), zinc (Zn) and manganese (Mn) are essential microelements with plant available fraction in soil, depending significantly on soil pH and soil organic matter (SOM), which is important for crop growth. The aim of this paper is to present the potential of mathematical models in order to predict the availability of microelements (Fe, Zn, Mn) in acidic and alkaline soils of eastern Croatia. The fundamental database for availability prediction contains results of 22,616 soil samples from eastern Croatia representing an area of 88,714.46 ha of arable land. The mandatory results include soil pH, SOM, available P and K, hydrolytic acidity, and carbonate content. Additional data sets, including supplementary results of total (extracted by aqua regia, AR) and available (extracted by ethylenediaminetetraacetate, EDTA) micronutrient fraction, were used for modelling of micronutrient availability and for final model validation. The modelling micronutrient available fraction was created in 3 steps: (1) regression models of total (AR) and available (EDTA) micronutrients (Fe, Zn, Mn) concentration based on analytical results of soil pH, SOM, AR and EDTA micronutrients fractions; (2) prediction of the available micronutrients fraction (EDTA) based on the soil pH and SOM; (3) model validation using new data set with analytical results of soil pH, SOM, AR and EDTA. The model predicts that moderate micronutrients availability could be expected on 48.45 % (42,972.25 out of 88,714.46 ha) of arable land on average for Fe, Zn and Mn. A high availability could be on 29,32 % (25.982 ha) of arable land on average, but a very significant difference was found among Fe (47,37 %), Mn (39,01 %) and Zn (1,57 %) arable land with high availability. The most important prediction is the one that claims insufficient availability of micronutrient could be expected on 19,579.87 ha in average, what is 22.26 % of arable land. But low Fe availability was predicted on only 2.79 % (2,479,3 ha), significantly more land (22.60 %, 20,035.40 ha) with low Mn availability and the highest percentage (41,4 %) of soil with insufficient Zn availability (36,764.91 out of 88,714.46 ha). The validation shows the highest model accuracy for Zn and the lowest for Fe availability prediction