The under-agarose chemotaxis assay method applied to glass microscope slides was optimised for canine polymorphonuclear granulocytes (PMN) with gelatin as protein supplement. Optimum chemotactic response occurred after approximately 135 min of incubation at 37°C with 5 x 10⁵ cells/well and zymosan-activated serum (ZAS) as chemoattractant. Chemotactic (C) and random migration (R) were stated separately for each dog. Storage of blood samples for 90 min before isolation of PMN did not influence the C or R values (p > 0.05). The preanalytical variation was investigated and the proportion of variance, expressed as a percentage, due to each source for the C and R values, respectively, was determined: dog, 92%, 96%; blood sample, 0.4%, 0%; interslide, 3%, 2% and intraslide, 3%, 3%. The mean chemotactic responsiveness against ZAS and ZAS diluted 1:1 with RPMI 1640 (dZAS) was determined using 15 laboratory beagle dogs: C(ZAS) = 0.67 mm (range: 0.25-2.33 mm), coefficient of variation (CV) = 24%, C(dZAS) = 0.47 mm (range: 0.18-2.12 mm), CV = 22%. Also, the mean random migration was determined: 0.11 mm (range. 0.05-0.34 mm), CV = 25%. Neither age nor sex had any significant influence on chemotactic or random migration (p > 0.05). The effect of heat-inactivated pooled canine serum (HI-serum) incorporated into agarose was examined and it was found that this had a significant stimulatory effect on both chemotactic and random migratory responses (p < 0.05). Therefore, agarose with HI-serum is not useful when investigating the chemotactic function.