As it is known, calcinating of cement clinker is accompanied by formation of outburst dust, containing considerable amounts of calcium and potassium. Since 1964, the kiln dust from the cement plant ＇＇Kunda Tehased＇ is widely used as a lime fertilizer in Estonia. Concentrations of macroelements and of six microelements in cement kiln dust, being used for chalking, were published in [1, 2]. At the present time, the growing ecological standards make necessary to widen the number of microelements to be controlled. In the present study concentrations of 30 microelements containing in cement kiln dust and their distribution between the dust precipitation stages were determined and some parameters were calculated. Two groups of samples were studied. The samples of C-1 series were taken from the screws under electrostatic precepitators of the 2nd, 3rd and 4th rotary kilns. The samples of the C-2 series were taken under the fields of the electrostatic precipitator of the 3rd rotary kiln; separately from the fields IV and V, totally from III and V, and from III to V. Using a high-sensitive analytical method, instrumental neutron activation analysis, and techniques described previously in [3, 4], concentrations of 30 microelements were determined (Table). The data on Zn and Co were compared with the corresponding ones reported earlier [1, 2], which were for Zn 74 [1] and 99 [2] g/tonne and for Co 2.15 [1] and 2.2 [2] g/tonne. Though the values to be compared have been defined by different methods and the interval between the studies is about 25 years, the numbers arc of the same order. This is the evidence, of the raw material composition being homogeneous and of the technological process being unchanged. To compare data on rare earth elements containing in kiln dust, they were normalized by hondrites (Fig. 1). Data on the composition of microelements in cement kiln outburst dusts were subjected to multidimensional statistical treatment. Based on factor analysis the generalized correlation matrix was obtained for 28 of the studied microelements. The correlation coefficients characterizing the degree of mutual stochastic interdependence of the element concentrations were calculated. The maximum number of significant correlations was found for Ta, namely 8. The analysis of the data obtained by means of the main component technique yielded the factor loading and the factor matrix F1, F2 and F3. A graphic representation of the results obtained in the planes of axes F1 - F2 and F1 - F3 is given in Figs. 2 and 3. In Fig. 2a one association of elements is distinguished, and in Fig. 2b three associations were already distinguished. In Fig. 3 both of the two diagrams contain clasters corresponding to initial groups C-1 and C-2.