Electronic counting of somatic cells in farm bulk tank milk, using chemical fat dispersion has been studied. The method used was the Pearson et al. modification of the original Tolle et al. procedure. It was found that fixation of the original sample with formalin was critical, and that with bulk tank milk this was best accomplished by the use of buffered formalin at 30–32 C for 24 h. No significant differences could be found, in samples kept up to 48 h at 4 C, between adding milk to buffered formalin and buffered formalin to milk. Good agreement was found between the DMSCC made on the day samples were taken, and electronic counts obtained following 24 and 48 h refrigeration storage before fixation, and 24 and 48 h following fixation after 24 h refrigeration. Correlation coefficients were + 0.876, + 0.924, and + 0.853, + 0.870, respectively. Replicate counts on both fresh bucket and bulk tank, and single counts on replicate samples from bulk tanks yielded coefficients of variation between 2.98% and 9.69%. The procedure shows promise of providing rapid, accurate somatic cell counts on large numbers of bulk tank milk samples.